[2026-03-25 20:35:09,647][mllm.models.large_language_model_local][INFO] - Initializing adapter 'agent_adapter': no initial weights provided or found; starting from scratch. [2026-03-25 20:35:10,392][mllm.models.adapter_training_wrapper][INFO] - Adapter 'agent_adapter': initialized with fresh weights (no initial weights found). [2026-03-25 20:35:10,398][mllm.models.large_language_model_local][INFO] - Initializing adapter 'critic_adapter': no initial weights provided or found; starting from scratch. [2026-03-25 20:35:11,101][mllm.models.adapter_training_wrapper][INFO] - Adapter 'critic_adapter': initialized with fresh weights (no initial weights found). [2026-03-25 20:37:28,172][__main__][INFO] - Starting iteration 0. [2026-03-25 20:37:28,177][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:37:28,178][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:37:41,584][__main__][INFO] - Number of regex retries in iteration 0: 0 [2026-03-25 20:37:41,585][__main__][INFO] - agents played in iteration 0 are Alice, Bob [2026-03-25 20:37:52,618][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 37.45%, Block Peak % of device VRAM: 18.68%, ΔTime: 00:00:00 [2026-03-25 20:37:52,662][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 37.45%, Block Peak % of device VRAM: 18.68%, ΔTime: 00:00:00 [2026-03-25 20:37:52,662][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:37:52,663][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:37:53,207][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:37:54,406][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:37:54,908][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:37:55,403][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:37:55,897][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:37:56,393][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:37:56,884][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:37:57,378][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:37:57,872][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:37:58,366][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:37:58,859][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:37:59,356][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:37:59,851][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:38:00,345][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:38:00,837][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:38:01,329][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:38:01,823][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:38:02,330][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:38:02,829][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:38:03,326][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:38:03,863][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:38:04,355][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:38:04,851][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:38:05,343][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:38:05,833][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:38:06,325][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:38:06,814][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:38:07,306][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:38:07,795][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:38:08,285][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:38:08,775][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:38:09,268][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:38:09,757][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:38:10,249][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:38:10,743][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:38:11,234][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:38:11,723][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:38:12,213][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:38:12,703][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:38:13,194][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:38:13,686][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:38:14,175][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:38:14,669][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:38:15,163][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:38:15,657][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:38:16,152][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:38:16,645][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:38:17,137][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:38:17,630][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:38:18,123][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:38:18,616][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:38:19,108][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:38:19,601][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:38:20,099][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:38:20,591][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:38:21,083][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:38:21,577][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:38:22,066][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:38:22,558][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:38:23,047][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:38:23,538][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:38:24,030][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:38:24,519][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:38:25,009][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:38:25,499][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:38:25,989][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:38:26,479][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:38:26,969][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:38:27,457][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:38:27,947][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:38:28,437][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:38:28,926][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:38:29,415][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:38:29,905][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:38:30,397][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:38:30,886][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:38:31,376][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:38:31,865][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:38:32,353][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:38:32,843][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:38:33,332][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:38:33,822][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:38:34,315][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:38:34,805][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:38:35,297][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:38:35,787][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:38:36,280][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:38:36,773][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:38:37,265][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:38:37,757][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:38:38,249][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:38:38,741][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:38:39,232][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:38:39,723][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:38:40,215][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:38:40,707][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:38:41,199][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:38:41,689][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:38:42,181][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:38:42,674][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:38:43,166][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:38:43,657][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:38:44,151][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:38:44,643][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:38:45,136][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:38:45,632][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:38:46,127][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:38:46,623][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:38:47,120][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:38:47,612][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:38:48,104][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:38:48,596][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:38:49,088][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:38:49,580][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:38:50,073][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:38:50,566][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:38:51,057][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:38:51,578][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:38:52,070][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:38:52,562][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:38:53,053][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:38:53,544][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:38:54,034][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:38:54,526][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:38:55,017][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:38:55,508][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:38:55,999][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:38:56,488][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:38:56,979][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19398 tokens. [2026-03-25 20:38:57,881][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 15.85%, Current % of VRAM taken: 53.30%, Block Peak % of device VRAM: 61.12%, ΔTime: 00:01:04 [2026-03-25 20:38:58,581][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:38:58,584][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:38:58,586][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:38:59,400][__main__][INFO] - Iteration 1 took 1m 31s (14.70% Gen, 84.41% Train). Generation: 13s, Training: 1m 17s. Estimated remaining time: 75h 56m 53s. Estimated total time: 76h 1m 10s. Time estimates for 10 more iterations: 15m 12s, 100 more iterations: 2h 32m 2s, 500 more iterations: 12h 40m 11s. [2026-03-25 20:38:59,402][__main__][INFO] - Starting iteration 1. [2026-03-25 20:38:59,805][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:38:59,805][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:39:00,435][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:39:12,011][__main__][INFO] - Number of regex retries in iteration 1: 1 [2026-03-25 20:39:12,012][__main__][INFO] - agents played in iteration 1 are Alice, Bob [2026-03-25 20:39:12,816][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 46.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:39:12,859][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 46.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:39:12,860][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:39:12,860][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:39:13,442][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:39:13,892][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:39:14,386][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:39:14,876][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:39:15,369][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:39:15,860][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:39:16,351][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:39:16,841][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:39:17,332][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:39:17,826][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:39:18,320][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:39:18,811][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:39:19,305][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:39:19,798][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:39:20,292][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:39:20,787][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:39:21,280][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:39:21,772][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:39:22,265][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:39:22,757][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:39:23,248][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:39:23,740][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:39:24,233][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:39:24,726][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:39:25,219][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:39:25,711][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:39:26,204][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:39:26,696][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:39:27,189][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:39:27,681][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:39:28,174][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:39:28,670][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:39:29,165][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:39:29,660][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:39:30,155][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:39:30,647][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:39:31,137][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:39:31,628][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:39:32,122][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:39:32,614][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:39:33,104][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:39:33,596][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:39:34,087][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:39:34,578][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:39:35,071][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:39:35,561][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:39:36,053][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:39:36,544][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:39:37,035][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:39:37,524][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:39:38,015][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:39:38,506][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:39:38,997][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:39:39,489][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:39:39,981][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:39:40,472][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:39:40,963][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:39:41,454][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:39:41,944][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:39:42,434][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:39:42,925][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:39:43,415][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:39:43,905][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:39:44,396][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:39:44,886][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:39:45,377][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:39:45,869][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:39:46,358][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:39:46,849][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:39:47,340][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:39:47,832][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:39:48,323][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:39:48,816][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:39:49,307][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:39:49,797][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:39:50,288][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:39:50,780][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:39:51,271][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:39:51,761][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:39:52,254][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:39:52,747][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:39:53,238][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:39:53,728][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:39:54,219][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:39:54,710][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:39:55,200][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:39:55,691][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:39:56,182][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:39:56,674][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:39:57,165][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:39:57,655][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:39:58,145][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:39:58,635][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:39:59,125][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:39:59,615][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:40:00,105][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:40:00,596][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:40:01,087][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:40:01,578][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:40:02,071][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:40:02,565][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:40:03,058][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:40:03,552][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:40:04,045][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:40:04,537][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:40:05,031][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:40:05,523][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:40:06,016][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:40:06,509][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:40:07,002][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:40:07,496][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:40:07,990][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:40:08,483][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:40:08,976][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:40:09,469][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:40:09,962][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:40:10,456][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:40:10,948][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:40:11,441][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:40:11,936][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:40:12,431][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:40:12,925][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:40:13,417][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:40:13,906][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:40:14,398][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:40:14,889][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:40:15,382][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:40:15,875][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:40:16,366][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19341 tokens. [2026-03-25 20:40:16,980][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.92%, Current % of VRAM taken: 60.28%, Block Peak % of device VRAM: 61.50%, ΔTime: 00:01:03 [2026-03-25 20:40:17,709][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:40:17,712][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:40:17,713][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:40:18,359][__main__][INFO] - Iteration 2 took 1m 18s (15.54% Gen, 83.64% Train). Generation: 12s, Training: 1m 5s. Estimated remaining time: 65h 22m 8s. Estimated total time: 65h 27m 44s. Time estimates for 10 more iterations: 13m 5s, 100 more iterations: 2h 10m 55s, 500 more iterations: 10h 54m 37s. [2026-03-25 20:40:18,361][__main__][INFO] - Starting iteration 2. [2026-03-25 20:40:18,764][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:40:18,765][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:40:30,746][__main__][INFO] - Number of regex retries in iteration 2: 0 [2026-03-25 20:40:30,747][__main__][INFO] - agents played in iteration 2 are Alice, Bob [2026-03-25 20:40:31,553][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.33%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:40:31,597][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.33%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:40:31,598][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:40:31,598][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:40:32,198][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:40:32,648][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:40:33,142][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:40:33,632][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:40:34,123][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:40:34,616][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:40:35,108][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:40:35,600][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:40:36,092][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:40:36,585][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:40:37,077][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:40:37,568][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:40:38,062][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:40:38,553][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:40:39,044][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:40:39,536][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:40:40,027][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:40:40,517][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:40:41,008][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:40:41,498][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:40:41,989][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:40:42,481][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:40:42,971][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:40:43,460][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:40:43,951][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:40:44,442][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:40:44,934][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:40:45,428][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:40:45,923][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:40:46,416][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:40:46,909][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:40:47,403][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:40:47,897][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:40:48,390][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:40:48,882][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:40:49,375][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:40:49,866][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:40:50,361][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:40:50,854][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:40:51,347][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:40:51,839][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:40:52,333][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:40:52,825][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:40:53,319][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:40:53,812][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:40:54,306][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:40:54,801][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:40:55,297][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:40:55,791][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:40:56,284][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:40:56,776][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:40:57,267][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:40:57,763][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:40:58,265][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:40:58,757][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:40:59,249][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:40:59,744][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:41:00,236][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:41:00,727][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:41:01,217][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:41:01,710][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:41:02,200][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:41:02,694][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:41:03,186][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:41:03,678][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:41:04,171][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:41:04,662][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:41:05,154][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:41:05,644][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:41:06,136][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:41:06,628][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:41:07,119][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:41:07,611][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:41:08,103][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:41:08,594][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:41:09,085][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:41:09,577][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:41:10,066][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:41:10,557][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:41:11,048][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:41:11,539][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:41:12,029][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:41:12,519][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:41:13,010][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:41:13,501][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:41:13,991][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:41:14,481][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:41:14,972][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:41:15,462][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:41:15,954][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:41:16,446][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:41:16,937][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:41:17,428][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:41:17,919][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:41:18,412][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:41:18,902][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:41:19,394][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:41:19,884][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:41:20,376][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:41:20,865][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:41:21,357][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:41:21,847][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:41:22,336][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:41:22,825][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:41:23,316][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:41:23,806][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:41:24,297][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:41:24,788][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:41:25,280][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:41:25,770][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:41:26,261][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:41:26,754][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:41:27,245][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:41:27,735][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:41:28,227][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:41:28,721][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:41:29,221][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:41:29,721][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:41:30,216][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:41:30,710][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:41:31,211][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:41:31,703][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:41:32,196][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:41:32,689][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:41:33,182][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:41:33,677][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:41:34,171][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:41:34,666][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:41:35,160][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19164 tokens. [2026-03-25 20:41:35,794][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.95%, Current % of VRAM taken: 60.31%, Block Peak % of device VRAM: 62.17%, ΔTime: 00:01:03 [2026-03-25 20:41:36,531][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:41:36,533][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:41:36,535][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:41:37,241][__main__][INFO] - Iteration 3 took 1m 18s (15.27% Gen, 83.83% Train). Generation: 11s, Training: 1m 5s. Estimated remaining time: 65h 16m 56s. Estimated total time: 65h 23m 51s. Time estimates for 10 more iterations: 13m 4s, 100 more iterations: 2h 10m 47s, 500 more iterations: 10h 53m 58s. [2026-03-25 20:41:37,243][__main__][INFO] - Starting iteration 3. [2026-03-25 20:41:37,658][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:41:37,659][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:41:49,954][__main__][INFO] - Number of regex retries in iteration 3: 0 [2026-03-25 20:41:49,955][__main__][INFO] - agents played in iteration 3 are Alice, Bob [2026-03-25 20:41:50,764][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.34%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:41:50,808][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.34%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:41:50,809][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:41:50,810][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:41:51,413][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:41:51,862][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:41:52,358][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:41:52,850][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:41:53,340][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:41:53,830][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:41:54,320][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:41:54,811][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:41:55,302][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:41:55,793][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:41:56,284][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:41:56,775][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:41:57,265][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:41:57,756][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:41:58,248][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:41:58,738][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:41:59,229][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:41:59,719][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:42:00,208][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:42:00,702][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:42:01,194][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:42:01,685][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:42:02,175][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:42:02,666][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:42:03,157][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:42:03,652][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:42:04,143][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:42:04,635][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:42:05,148][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:42:05,640][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:42:06,132][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:42:06,623][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:42:07,115][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:42:07,607][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:42:08,099][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:42:08,590][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:42:09,082][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:42:09,574][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:42:10,068][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:42:10,559][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:42:11,051][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:42:11,541][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:42:12,031][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:42:12,524][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:42:13,017][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:42:13,510][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:42:14,003][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:42:14,497][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:42:14,991][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:42:15,484][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:42:15,977][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:42:16,471][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:42:16,964][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:42:17,456][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:42:17,949][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:42:18,442][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:42:18,935][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:42:19,430][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:42:19,924][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:42:20,418][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:42:20,912][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:42:21,411][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:42:21,907][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:42:22,403][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:42:22,899][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:42:23,391][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:42:23,882][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:42:24,375][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:42:24,864][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:42:25,356][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:42:25,847][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:42:26,337][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:42:26,827][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:42:27,318][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:42:27,810][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:42:28,301][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:42:28,792][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:42:29,283][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:42:29,775][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:42:30,266][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:42:30,757][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:42:31,248][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:42:31,739][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:42:32,231][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:42:32,723][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:42:33,215][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:42:33,705][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:42:34,196][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:42:34,687][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:42:35,179][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:42:35,672][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:42:36,164][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:42:36,656][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:42:37,147][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:42:37,641][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:42:38,134][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:42:38,627][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:42:39,118][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:42:39,609][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:42:40,100][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:42:40,592][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:42:41,084][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:42:41,576][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:42:42,067][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:42:42,558][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:42:43,049][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:42:43,539][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:42:44,030][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:42:44,520][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:42:45,011][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:42:45,502][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:42:45,993][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:42:46,484][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:42:46,976][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:42:47,466][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:42:47,957][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:42:48,451][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:42:48,942][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:42:49,434][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:42:49,924][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:42:50,415][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:42:50,906][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:42:51,398][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:42:51,893][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:42:52,385][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:42:52,878][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:42:53,371][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:42:53,864][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:42:54,356][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19249 tokens. [2026-03-25 20:42:54,982][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.99%, Current % of VRAM taken: 60.35%, Block Peak % of device VRAM: 61.61%, ΔTime: 00:01:03 [2026-03-25 20:42:55,744][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:42:55,746][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:42:55,748][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:42:56,470][__main__][INFO] - Iteration 4 took 1m 18s (15.60% Gen, 83.48% Train). Generation: 12s, Training: 1m 5s. Estimated remaining time: 65h 32m 25s. Estimated total time: 65h 40m 39s. Time estimates for 10 more iterations: 13m 8s, 100 more iterations: 2h 11m 21s, 500 more iterations: 10h 56m 46s. [2026-03-25 20:42:56,473][__main__][INFO] - Starting iteration 4. [2026-03-25 20:42:56,884][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:42:56,885][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:43:09,778][__main__][INFO] - Number of regex retries in iteration 4: 0 [2026-03-25 20:43:09,779][__main__][INFO] - agents played in iteration 4 are Alice, Bob [2026-03-25 20:43:10,614][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:43:10,658][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:43:10,659][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:43:10,659][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:43:11,273][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:43:11,723][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:43:12,223][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:43:12,718][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:43:13,210][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:43:13,703][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:43:14,201][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:43:14,694][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:43:15,188][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:43:15,683][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:43:16,178][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:43:16,670][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:43:17,162][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:43:17,656][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:43:18,150][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:43:18,644][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:43:19,136][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:43:19,631][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:43:20,124][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:43:20,619][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:43:21,115][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:43:21,609][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:43:22,102][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:43:22,593][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:43:23,083][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:43:23,575][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:43:24,067][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:43:24,558][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:43:25,050][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:43:25,540][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:43:26,032][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:43:26,523][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:43:27,015][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:43:27,506][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:43:27,997][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:43:28,489][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:43:28,980][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:43:29,470][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:43:29,962][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:43:30,453][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:43:30,943][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:43:31,434][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:43:31,923][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:43:32,415][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:43:32,905][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:43:33,396][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:43:33,887][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:43:34,379][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:43:34,870][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:43:35,360][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:43:35,852][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:43:36,341][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:43:36,832][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:43:37,322][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:43:37,814][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:43:38,303][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:43:38,793][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:43:39,286][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:43:39,778][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:43:40,270][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:43:40,761][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:43:41,254][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:43:41,745][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:43:42,237][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:43:42,729][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:43:43,220][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:43:43,712][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:43:44,205][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:43:44,696][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:43:45,188][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:43:45,680][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:43:46,171][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:43:46,664][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:43:47,155][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:43:47,646][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:43:48,138][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:43:48,629][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:43:49,119][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:43:49,610][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:43:50,100][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:43:50,591][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:43:51,083][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:43:51,576][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:43:52,068][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:43:52,559][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:43:53,050][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:43:53,540][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:43:54,030][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:43:54,525][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:43:55,019][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:43:55,512][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:43:56,004][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:43:56,499][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:43:56,994][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:43:57,488][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:43:57,981][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:43:58,475][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:43:58,968][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:43:59,460][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:43:59,952][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:44:00,444][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:44:00,936][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:44:01,429][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:44:01,921][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:44:02,413][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:44:02,905][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:44:03,398][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:44:03,892][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:44:04,385][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:44:04,879][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:44:05,371][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:44:05,861][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:44:06,352][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:44:06,842][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:44:07,332][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:44:07,822][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:44:08,312][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:44:08,802][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:44:09,293][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:44:09,782][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:44:10,273][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:44:10,764][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:44:11,256][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:44:11,749][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:44:12,241][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:44:12,734][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:44:13,225][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:44:13,717][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:44:14,210][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19228 tokens. [2026-03-25 20:44:14,840][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.00%, Current % of VRAM taken: 60.36%, Block Peak % of device VRAM: 61.60%, ΔTime: 00:01:03 [2026-03-25 20:44:15,572][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:44:15,575][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:44:15,576][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:44:16,217][__main__][INFO] - Iteration 5 took 1m 19s (16.25% Gen, 82.94% Train). Generation: 12s, Training: 1m 5s. Estimated remaining time: 65h 57m 6s. Estimated total time: 66h 6m 40s. Time estimates for 10 more iterations: 13m 13s, 100 more iterations: 2h 12m 13s, 500 more iterations: 11h 1m 6s. [2026-03-25 20:44:16,219][__main__][INFO] - Starting iteration 5. [2026-03-25 20:44:16,619][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:44:16,620][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:44:30,141][__main__][INFO] - Number of regex retries in iteration 5: 0 [2026-03-25 20:44:30,142][__main__][INFO] - agents played in iteration 5 are Alice, Bob [2026-03-25 20:44:31,157][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:44:31,202][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:44:31,203][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:44:31,203][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:44:31,815][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:44:32,266][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:44:32,765][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:44:33,263][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:44:33,758][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:44:34,253][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:44:34,746][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:44:35,239][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:44:35,732][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:44:36,224][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:44:36,716][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:44:37,208][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:44:37,704][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:44:38,197][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:44:38,689][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:44:39,204][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:44:39,696][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:44:40,189][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:44:40,682][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:44:41,174][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:44:41,668][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:44:42,160][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:44:42,656][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:44:43,149][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:44:43,641][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:44:44,133][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:44:44,625][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:44:45,117][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:44:45,610][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:44:46,102][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:44:46,594][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:44:47,084][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:44:47,576][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:44:48,069][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:44:48,563][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:44:49,056][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:44:49,547][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:44:50,040][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:44:50,533][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:44:51,025][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:44:51,518][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:44:52,010][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:44:52,502][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:44:52,995][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:44:53,487][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:44:53,981][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:44:54,474][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:44:54,967][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:44:55,462][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:44:55,953][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:44:56,444][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:44:56,936][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:44:57,427][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:44:57,918][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:44:58,409][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:44:58,899][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:44:59,390][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:44:59,884][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:45:00,376][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:45:00,866][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:45:01,358][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:45:01,849][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:45:02,340][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:45:02,831][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:45:03,322][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:45:03,814][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:45:04,305][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:45:04,797][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:45:05,288][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:45:05,780][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:45:06,275][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:45:06,771][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:45:07,265][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:45:07,761][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:45:08,257][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:45:08,751][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:45:09,247][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:45:09,740][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:45:10,235][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:45:10,731][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:45:11,225][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:45:11,719][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:45:12,213][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:45:12,707][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:45:13,202][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:45:13,697][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:45:14,193][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:45:14,688][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:45:15,181][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:45:15,676][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:45:16,172][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:45:16,667][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:45:17,161][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:45:17,652][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:45:18,143][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:45:18,635][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:45:19,126][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:45:19,618][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:45:20,110][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:45:20,601][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:45:21,092][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:45:21,586][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:45:22,078][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:45:22,569][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:45:23,061][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:45:23,555][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:45:24,048][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:45:24,539][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:45:25,031][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:45:25,523][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:45:26,015][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:45:26,507][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:45:27,000][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:45:27,492][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:45:27,984][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:45:28,477][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:45:28,969][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:45:29,459][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:45:29,949][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:45:30,440][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:45:30,930][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:45:31,422][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:45:31,913][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:45:32,403][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:45:32,894][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:45:33,385][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:45:33,876][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:45:34,370][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:45:34,863][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19357 tokens. [2026-03-25 20:45:35,490][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.96%, Current % of VRAM taken: 60.32%, Block Peak % of device VRAM: 61.68%, ΔTime: 00:01:03 [2026-03-25 20:45:36,285][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:45:36,287][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:45:36,288][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:45:37,129][__main__][INFO] - Iteration 6 took 1m 20s (16.80% Gen, 82.16% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 54m 35s. Estimated total time: 67h 5m 30s. Time estimates for 10 more iterations: 13m 25s, 100 more iterations: 2h 14m 11s, 500 more iterations: 11h 10m 55s. [2026-03-25 20:45:37,131][__main__][INFO] - Starting iteration 6. [2026-03-25 20:45:37,533][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:45:37,534][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:45:49,081][__main__][INFO] - Number of regex retries in iteration 6: 0 [2026-03-25 20:45:49,082][__main__][INFO] - agents played in iteration 6 are Alice, Bob [2026-03-25 20:45:50,079][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:45:50,124][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:45:50,125][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:45:50,125][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:45:50,736][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:45:51,187][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:45:51,686][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:45:52,184][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:45:52,679][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:45:53,176][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:45:53,670][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:45:54,165][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:45:54,660][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:45:55,155][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:45:55,650][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:45:56,148][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:45:56,644][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:45:57,138][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:45:57,634][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:45:58,128][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:45:58,621][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:45:59,117][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:45:59,611][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:46:00,106][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:46:00,601][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:46:01,095][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:46:01,587][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:46:02,079][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:46:02,571][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:46:03,064][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:46:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:46:04,050][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:46:04,542][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:46:05,036][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:46:05,538][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:46:06,031][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:46:06,524][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:46:07,016][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:46:07,507][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:46:07,998][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:46:08,490][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:46:08,982][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:46:09,476][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:46:09,967][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:46:10,460][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:46:10,951][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:46:11,442][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:46:11,933][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:46:12,423][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:46:12,915][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:46:13,408][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:46:13,899][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:46:14,391][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:46:14,882][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:46:15,379][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:46:15,876][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:46:16,368][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:46:16,861][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:46:17,354][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:46:17,845][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:46:18,340][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:46:18,831][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:46:19,322][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:46:19,813][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:46:20,304][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:46:20,796][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:46:21,287][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:46:21,779][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:46:22,269][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:46:22,761][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:46:23,253][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:46:23,745][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:46:24,239][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:46:24,730][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:46:25,222][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:46:25,715][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:46:26,208][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:46:26,699][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:46:27,192][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:46:27,684][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:46:28,178][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:46:28,670][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:46:29,162][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:46:29,654][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:46:30,144][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:46:30,636][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:46:31,128][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:46:31,620][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:46:32,113][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:46:32,604][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:46:33,099][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:46:33,593][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:46:34,087][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:46:34,582][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:46:35,078][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:46:35,574][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:46:36,069][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:46:36,564][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:46:37,063][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:46:37,557][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:46:38,055][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:46:38,550][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:46:39,043][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:46:39,537][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:46:40,031][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:46:40,525][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:46:41,020][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:46:41,514][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:46:42,010][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:46:42,507][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:46:43,002][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:46:43,498][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:46:43,990][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:46:44,481][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:46:44,975][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:46:45,466][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:46:45,958][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:46:46,448][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:46:46,941][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:46:47,435][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:46:47,927][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:46:48,422][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:46:48,915][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:46:49,409][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:46:49,902][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:46:50,394][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:46:50,885][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:46:51,377][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:46:51,867][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:46:52,358][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:46:52,848][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:46:53,339][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:46:53,829][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19246 tokens. [2026-03-25 20:46:54,476][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.92%, Current % of VRAM taken: 60.28%, Block Peak % of device VRAM: 61.83%, ΔTime: 00:01:03 [2026-03-25 20:46:55,205][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:46:55,208][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:46:55,209][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:46:55,900][__main__][INFO] - Iteration 7 took 1m 18s (14.74% Gen, 84.38% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 65h 6m 9s. Estimated total time: 65h 18m 23s. Time estimates for 10 more iterations: 13m 3s, 100 more iterations: 2h 10m 36s, 500 more iterations: 10h 53m 3s. [2026-03-25 20:46:55,903][__main__][INFO] - Starting iteration 7. [2026-03-25 20:46:56,303][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:46:56,304][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:46:56,929][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:46:56,934][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:47:09,000][__main__][INFO] - Number of regex retries in iteration 7: 2 [2026-03-25 20:47:09,001][__main__][INFO] - agents played in iteration 7 are Alice, Bob [2026-03-25 20:47:09,816][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:47:09,861][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:47:09,861][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:47:09,862][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:47:10,480][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:47:10,930][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:47:11,427][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:47:11,919][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:47:12,411][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:47:12,903][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:47:13,394][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:47:13,886][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:47:14,379][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:47:14,872][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:47:15,364][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:47:15,859][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:47:16,359][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:47:16,856][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:47:17,352][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:47:17,847][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:47:18,343][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:47:18,842][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:47:19,339][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:47:19,836][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:47:20,331][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:47:20,827][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:47:21,323][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:47:21,819][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:47:22,313][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:47:22,810][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:47:23,304][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:47:23,799][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:47:24,293][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:47:24,788][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:47:25,284][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:47:25,780][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:47:26,278][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:47:26,777][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:47:27,296][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:47:27,790][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:47:28,284][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:47:28,778][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:47:29,270][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:47:29,762][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:47:30,257][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:47:30,749][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:47:31,240][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:47:31,730][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:47:32,223][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:47:32,717][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:47:33,208][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:47:33,700][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:47:34,191][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:47:34,682][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:47:35,174][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:47:35,665][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:47:36,158][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:47:36,648][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:47:37,139][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:47:37,629][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:47:38,120][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:47:38,615][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:47:39,109][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:47:39,605][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:47:40,098][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:47:40,590][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:47:41,083][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:47:41,575][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:47:42,068][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:47:42,560][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:47:43,051][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:47:43,543][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:47:44,034][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:47:44,527][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:47:45,018][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:47:45,512][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:47:46,007][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:47:46,500][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:47:46,994][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:47:47,487][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:47:47,981][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:47:48,479][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:47:48,973][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:47:49,465][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:47:49,958][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:47:50,451][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:47:50,945][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:47:51,438][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:47:51,932][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:47:52,424][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:47:52,917][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:47:53,410][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:47:53,903][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:47:54,395][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:47:54,887][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:47:55,380][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:47:55,873][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:47:56,365][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:47:56,858][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:47:57,349][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:47:57,843][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:47:58,338][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:47:58,832][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:47:59,328][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:47:59,823][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:48:00,319][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:48:00,814][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:48:01,308][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:48:01,805][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:48:02,300][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:48:02,794][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:48:03,288][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:48:03,782][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:48:04,279][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:48:04,773][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:48:05,267][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:48:05,761][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:48:06,256][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:48:06,749][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:48:07,243][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:48:07,736][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:48:08,230][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:48:08,724][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:48:09,219][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:48:09,714][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:48:10,207][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:48:10,700][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:48:11,194][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:48:11,687][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:48:12,180][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:48:12,674][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:48:13,167][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:48:13,661][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19442 tokens. [2026-03-25 20:48:14,307][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.02%, Current % of VRAM taken: 60.38%, Block Peak % of device VRAM: 61.65%, ΔTime: 00:01:03 [2026-03-25 20:48:15,045][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:48:15,047][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:48:15,049][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:48:15,703][__main__][INFO] - Iteration 8 took 1m 19s (15.99% Gen, 83.18% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 56m 27s. Estimated total time: 66h 10m 1s. Time estimates for 10 more iterations: 13m 14s, 100 more iterations: 2h 12m 20s, 500 more iterations: 11h 1m 40s. [2026-03-25 20:48:15,705][__main__][INFO] - Starting iteration 8. [2026-03-25 20:48:16,102][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:48:16,103][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:48:19,888][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:48:21,988][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:48:23,073][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls Given the per-item values: - You value hats at 10, books at 1, and balls at 1. - Bob values hats at 1, books at 10, and balls at 1. Since you have a much higher value for hats compared to Bob, proposing to take all 10 hats ensures you get the maximum possible value from this round. Bob is more interested in books, so by not proposing to take any books or balls, you avoid splitting potentially less valuable items. did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:48:29,409][__main__][INFO] - Number of regex retries in iteration 8: 3 [2026-03-25 20:48:29,410][__main__][INFO] - agents played in iteration 8 are Alice, Bob [2026-03-25 20:48:30,220][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.38%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:48:30,265][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.38%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:48:30,266][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:48:30,266][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:48:30,873][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:48:31,322][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:48:31,818][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:48:32,314][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:48:32,806][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:48:33,298][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:48:33,789][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:48:34,281][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:48:34,771][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:48:35,264][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:48:35,758][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:48:36,252][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:48:36,745][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:48:37,238][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:48:37,732][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:48:38,225][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:48:38,718][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:48:39,208][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:48:39,700][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:48:40,190][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:48:40,681][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:48:41,172][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:48:41,663][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:48:42,158][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:48:42,654][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:48:43,149][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:48:43,645][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:48:44,141][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:48:44,637][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:48:45,130][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:48:45,625][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:48:46,119][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:48:46,614][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:48:47,108][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:48:47,602][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:48:48,095][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:48:48,590][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:48:49,083][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:48:49,578][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:48:50,072][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:48:50,565][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:48:51,061][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:48:51,557][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:48:52,079][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:48:52,581][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:48:53,079][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:48:53,574][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:48:54,068][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:48:54,560][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:48:55,052][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:48:55,543][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:48:56,036][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:48:56,529][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:48:57,021][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:48:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:48:58,008][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:48:58,503][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:48:59,001][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:48:59,495][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:48:59,990][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:49:00,482][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:49:00,978][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:49:01,470][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:49:01,962][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:49:02,456][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:49:02,948][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:49:03,440][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:49:03,932][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:49:04,423][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:49:04,915][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:49:05,406][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:49:05,898][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:49:06,389][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:49:06,882][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:49:07,376][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:49:07,869][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:49:08,362][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:49:08,855][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:49:09,347][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:49:09,840][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:49:10,334][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:49:10,824][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:49:11,316][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:49:11,807][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:49:12,299][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:49:12,789][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:49:13,281][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:49:13,771][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:49:14,263][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:49:14,756][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:49:15,248][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:49:15,743][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:49:16,237][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:49:16,728][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:49:17,221][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:49:17,711][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:49:18,206][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:49:18,702][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:49:19,194][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:49:19,685][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:49:20,178][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:49:20,667][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:49:21,159][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:49:21,651][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:49:22,142][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:49:22,635][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:49:23,127][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:49:23,620][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:49:24,114][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:49:24,608][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:49:25,103][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:49:25,600][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:49:26,093][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:49:26,588][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:49:27,081][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:49:27,578][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:49:28,073][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:49:28,566][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:49:29,061][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:49:29,560][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:49:30,057][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:49:30,556][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:49:31,052][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:49:31,547][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:49:32,042][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:49:32,537][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:49:33,032][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:49:33,527][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:49:34,023][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19395 tokens. [2026-03-25 20:49:34,681][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.04%, Current % of VRAM taken: 60.40%, Block Peak % of device VRAM: 61.73%, ΔTime: 00:01:03 [2026-03-25 20:49:35,424][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:49:35,426][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:49:35,428][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:49:36,276][__main__][INFO] - Iteration 9 took 1m 20s (16.60% Gen, 82.34% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 33m 49s. Estimated total time: 66h 48m 43s. Time estimates for 10 more iterations: 13m 21s, 100 more iterations: 2h 13m 37s, 500 more iterations: 11h 8m 7s. [2026-03-25 20:49:36,278][__main__][INFO] - Starting iteration 9. [2026-03-25 20:49:36,676][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:49:36,677][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:49:48,801][__main__][INFO] - Number of regex retries in iteration 9: 0 [2026-03-25 20:49:48,802][__main__][INFO] - agents played in iteration 9 are Alice, Bob [2026-03-25 20:49:49,622][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:49:49,667][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:49:49,668][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:49:49,668][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:49:50,277][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:49:50,729][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:49:51,227][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:49:51,720][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:49:52,212][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:49:52,706][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:49:53,199][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:49:53,690][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:49:54,183][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:49:54,675][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:49:55,167][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:49:55,662][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:49:56,156][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:49:56,647][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:49:57,139][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:49:57,634][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:49:58,125][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:49:58,619][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:49:59,111][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:49:59,605][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:50:00,098][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:50:00,590][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:50:01,083][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:50:01,576][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:50:02,068][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:50:02,561][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:50:03,053][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:50:03,548][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:50:04,040][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:50:04,536][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:50:05,030][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:50:05,522][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:50:06,017][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:50:06,509][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:50:07,002][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:50:07,494][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:50:07,987][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:50:08,479][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:50:08,975][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:50:09,470][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:50:09,963][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:50:10,458][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:50:10,952][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:50:11,446][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:50:11,941][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:50:12,434][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:50:12,929][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:50:13,426][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:50:13,923][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:50:14,419][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:50:14,913][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:50:15,408][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:50:15,903][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:50:16,398][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:50:16,891][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:50:17,386][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:50:17,880][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:50:18,378][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:50:18,876][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:50:19,374][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:50:19,871][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:50:20,365][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:50:20,858][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:50:21,351][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:50:21,844][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:50:22,337][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:50:22,830][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:50:23,324][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:50:23,817][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:50:24,311][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:50:24,805][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:50:25,298][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:50:25,791][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:50:26,283][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:50:26,775][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:50:27,266][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:50:27,759][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:50:28,251][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:50:28,743][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:50:29,237][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:50:29,729][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:50:30,221][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:50:30,713][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:50:31,209][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:50:31,701][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:50:32,192][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:50:32,684][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:50:33,178][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:50:33,670][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:50:34,163][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:50:34,657][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:50:35,151][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:50:35,644][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:50:36,138][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:50:36,631][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:50:37,124][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:50:37,619][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:50:38,110][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:50:38,602][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:50:39,095][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:50:39,588][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:50:40,081][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:50:40,572][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:50:41,064][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:50:41,555][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:50:42,047][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:50:42,539][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:50:43,032][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:50:43,524][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:50:44,016][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:50:44,508][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:50:45,002][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:50:45,499][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:50:45,993][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:50:46,485][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:50:46,978][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:50:47,469][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:50:47,962][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:50:48,456][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:50:48,949][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:50:49,439][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:50:49,932][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:50:50,425][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:50:50,919][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:50:51,412][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:50:51,905][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:50:52,398][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:50:52,893][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:50:53,389][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19342 tokens. [2026-03-25 20:50:54,053][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.97%, Current % of VRAM taken: 60.33%, Block Peak % of device VRAM: 61.59%, ΔTime: 00:01:03 [2026-03-25 20:50:54,791][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:50:54,794][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:50:54,796][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:50:55,575][__main__][INFO] - Iteration 10 took 1m 18s (15.37% Gen, 83.64% Train). Generation: 12s, Training: 1m 5s. Estimated remaining time: 65h 28m 46s. Estimated total time: 65h 45m 0s. Time estimates for 10 more iterations: 13m 9s, 100 more iterations: 2h 11m 30s, 500 more iterations: 10h 57m 30s. [2026-03-25 20:50:55,578][__main__][INFO] - Starting iteration 10. [2026-03-25 20:50:55,989][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:50:55,990][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:50:56,801][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:51:09,806][__main__][INFO] - Number of regex retries in iteration 10: 1 [2026-03-25 20:51:09,807][__main__][INFO] - agents played in iteration 10 are Alice, Bob [2026-03-25 20:51:10,629][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:51:10,674][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:51:10,674][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:51:10,675][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:51:11,278][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:51:11,728][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:51:12,224][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:51:12,720][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:51:13,213][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:51:13,705][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:51:14,200][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:51:14,695][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:51:15,189][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:51:15,682][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:51:16,175][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:51:16,669][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:51:17,162][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:51:17,655][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:51:18,146][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:51:18,641][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:51:19,134][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:51:19,626][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:51:20,119][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:51:20,611][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:51:21,104][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:51:21,596][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:51:22,089][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:51:22,580][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:51:23,072][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:51:23,564][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:51:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:51:24,549][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:51:25,042][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:51:25,536][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:51:26,029][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:51:26,523][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:51:27,017][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:51:27,510][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:51:28,004][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:51:28,498][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:51:28,993][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:51:29,491][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:51:29,985][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:51:30,478][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:51:30,972][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:51:31,465][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:51:31,958][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:51:32,450][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:51:32,943][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:51:33,435][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:51:33,927][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:51:34,420][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:51:34,913][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:51:35,405][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:51:35,898][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:51:36,391][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:51:36,885][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:51:37,380][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:51:37,873][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:51:38,365][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:51:38,858][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:51:39,350][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:51:39,842][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:51:40,333][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:51:40,827][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:51:41,319][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:51:41,814][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:51:42,308][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:51:42,803][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:51:43,298][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:51:43,794][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:51:44,289][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:51:44,785][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:51:45,281][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:51:45,777][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:51:46,272][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:51:46,767][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:51:47,262][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:51:47,756][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:51:48,252][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:51:48,746][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:51:49,241][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:51:49,734][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:51:50,229][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:51:50,723][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:51:51,221][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:51:51,718][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:51:52,213][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:51:52,708][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:51:53,201][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:51:53,694][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:51:54,185][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:51:54,678][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:51:55,171][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:51:55,664][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:51:56,157][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:51:56,649][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:51:57,141][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:51:57,636][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:51:58,129][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:51:58,622][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:51:59,116][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:51:59,610][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:52:00,102][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:52:00,596][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:52:01,091][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:52:01,584][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:52:02,081][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:52:02,576][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:52:03,069][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:52:03,563][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:52:04,055][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:52:04,548][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:52:05,040][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:52:05,533][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:52:06,026][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:52:06,520][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:52:07,013][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:52:07,505][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:52:07,998][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:52:08,490][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:52:08,984][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:52:09,476][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:52:09,969][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:52:10,461][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:52:10,953][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:52:11,445][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:52:11,935][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:52:12,428][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:52:12,921][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:52:13,413][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:52:13,906][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:52:14,398][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19272 tokens. [2026-03-25 20:52:15,050][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.95%, Current % of VRAM taken: 60.31%, Block Peak % of device VRAM: 61.64%, ΔTime: 00:01:03 [2026-03-25 20:52:15,786][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:52:15,788][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:52:15,790][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:52:16,441][__main__][INFO] - Iteration 11 took 1m 20s (17.17% Gen, 82.01% Train). Generation: 13s, Training: 1m 5s. Estimated remaining time: 66h 45m 4s. Estimated total time: 67h 2m 38s. Time estimates for 10 more iterations: 13m 24s, 100 more iterations: 2h 14m 5s, 500 more iterations: 11h 10m 26s. [2026-03-25 20:52:16,444][__main__][INFO] - Starting iteration 11. [2026-03-25 20:52:16,844][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:52:16,845][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:52:17,489][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:52:30,142][__main__][INFO] - Number of regex retries in iteration 11: 1 [2026-03-25 20:52:30,143][__main__][INFO] - agents played in iteration 11 are Alice, Bob [2026-03-25 20:52:31,168][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.35%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:52:31,213][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.35%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:52:31,214][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:52:31,215][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:52:31,834][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:52:32,287][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:52:32,788][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:52:33,285][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:52:33,783][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:52:34,278][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:52:34,774][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:52:35,269][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:52:35,766][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:52:36,262][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:52:36,757][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:52:37,252][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:52:37,748][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:52:38,243][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:52:38,739][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:52:39,231][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:52:39,723][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:52:40,215][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:52:40,707][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:52:41,201][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:52:41,699][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:52:42,194][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:52:42,686][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:52:43,181][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:52:43,674][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:52:44,167][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:52:44,661][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:52:45,156][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:52:45,651][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:52:46,145][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:52:46,638][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:52:47,130][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:52:47,623][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:52:48,115][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:52:48,608][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:52:49,100][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:52:49,592][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:52:50,085][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:52:50,578][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:52:51,070][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:52:51,562][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:52:52,053][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:52:52,544][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:52:53,034][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:52:53,526][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:52:54,018][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:52:54,508][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:52:55,000][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:52:55,492][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:52:55,985][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:52:56,479][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:52:56,972][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:52:57,467][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:52:57,960][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:52:58,454][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:52:58,948][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:52:59,442][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:52:59,935][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:53:00,428][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:53:00,920][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:53:01,414][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:53:01,909][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:53:02,402][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:53:02,897][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:53:03,391][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:53:03,885][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:53:04,379][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:53:04,872][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:53:05,365][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:53:05,858][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:53:06,351][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:53:06,843][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:53:07,337][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:53:07,830][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:53:08,322][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:53:08,814][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:53:09,306][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:53:09,799][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:53:10,292][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:53:10,784][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:53:11,279][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:53:11,773][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:53:12,268][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:53:12,764][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:53:13,261][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:53:13,756][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:53:14,251][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:53:14,746][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:53:15,241][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:53:15,736][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:53:16,231][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:53:16,727][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:53:17,222][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:53:17,717][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:53:18,212][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:53:18,707][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:53:19,203][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:53:19,697][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:53:20,190][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:53:20,683][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:53:21,180][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:53:21,676][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:53:22,172][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:53:22,665][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:53:23,158][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:53:23,649][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:53:24,141][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:53:24,632][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:53:25,124][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:53:25,617][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:53:26,108][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:53:26,600][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:53:27,091][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:53:27,584][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:53:28,078][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:53:28,570][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:53:29,063][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:53:29,557][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:53:30,050][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:53:30,543][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:53:31,038][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:53:31,531][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:53:32,024][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:53:32,518][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:53:33,012][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:53:33,506][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:53:33,999][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:53:34,492][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:53:34,986][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19325 tokens. [2026-03-25 20:53:35,627][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.04%, Current % of VRAM taken: 60.40%, Block Peak % of device VRAM: 61.74%, ΔTime: 00:01:03 [2026-03-25 20:53:36,438][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:53:36,441][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:53:36,442][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:53:37,167][__main__][INFO] - Iteration 12 took 1m 20s (16.56% Gen, 82.54% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 37m 15s. Estimated total time: 66h 56m 11s. Time estimates for 10 more iterations: 13m 23s, 100 more iterations: 2h 13m 52s, 500 more iterations: 11h 9m 21s. [2026-03-25 20:53:37,170][__main__][INFO] - Starting iteration 12. [2026-03-25 20:53:37,572][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:53:37,573][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:53:49,647][__main__][INFO] - Number of regex retries in iteration 12: 0 [2026-03-25 20:53:49,648][__main__][INFO] - agents played in iteration 12 are Alice, Bob [2026-03-25 20:53:50,463][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:53:50,508][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:53:50,509][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:53:50,510][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:53:51,120][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:53:51,569][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:53:52,067][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:53:52,559][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:53:53,051][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:53:53,544][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:53:54,039][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:53:54,530][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:53:55,025][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:53:55,520][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:53:56,014][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:53:56,512][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:53:57,006][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:53:57,501][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:53:57,995][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:53:58,489][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:53:58,984][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:53:59,481][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:53:59,999][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:54:00,495][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:54:00,990][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:54:01,484][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:54:01,982][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:54:02,477][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:54:02,972][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:54:03,468][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:54:03,964][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:54:04,461][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:54:04,960][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:54:05,457][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:54:05,951][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:54:06,445][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:54:06,941][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:54:07,433][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:54:07,925][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:54:08,415][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:54:08,906][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:54:09,398][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:54:09,890][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:54:10,383][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:54:10,876][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:54:11,373][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:54:11,866][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:54:12,361][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:54:12,855][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:54:13,347][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:54:13,841][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:54:14,335][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:54:14,827][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:54:15,319][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:54:15,812][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:54:16,305][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:54:16,797][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:54:17,290][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:54:17,783][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:54:18,277][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:54:18,770][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:54:19,265][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:54:19,761][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:54:20,254][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:54:20,746][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:54:21,239][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:54:21,733][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:54:22,227][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:54:22,721][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:54:23,212][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:54:23,704][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:54:24,196][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:54:24,688][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:54:25,180][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:54:25,672][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:54:26,164][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:54:26,656][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:54:27,147][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:54:27,639][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:54:28,133][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:54:28,625][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:54:29,116][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:54:29,608][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:54:30,100][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:54:30,593][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:54:31,086][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:54:31,580][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:54:32,074][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:54:32,567][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:54:33,059][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:54:33,551][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:54:34,043][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:54:34,536][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:54:35,031][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:54:35,524][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:54:36,018][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:54:36,512][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:54:37,004][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:54:37,498][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:54:37,992][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:54:38,488][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:54:38,983][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:54:39,477][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:54:39,971][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:54:40,464][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:54:40,957][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:54:41,452][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:54:41,945][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:54:42,440][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:54:42,938][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:54:43,434][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:54:43,929][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:54:44,424][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:54:44,920][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:54:45,415][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:54:45,911][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:54:46,407][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:54:46,901][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:54:47,395][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:54:47,891][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:54:48,387][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:54:48,884][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:54:49,380][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:54:49,873][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:54:50,366][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:54:50,859][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:54:51,352][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:54:51,847][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:54:52,340][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:54:52,833][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:54:53,327][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:54:53,820][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:54:54,314][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19256 tokens. [2026-03-25 20:54:54,954][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.00%, Current % of VRAM taken: 60.36%, Block Peak % of device VRAM: 61.67%, ΔTime: 00:01:03 [2026-03-25 20:54:55,690][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:54:55,693][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:54:55,694][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:54:56,394][__main__][INFO] - Iteration 13 took 1m 18s (15.32% Gen, 83.79% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 20m 51s. Estimated total time: 65h 41m 6s. Time estimates for 10 more iterations: 13m 8s, 100 more iterations: 2h 11m 22s, 500 more iterations: 10h 56m 51s. [2026-03-25 20:54:56,396][__main__][INFO] - Starting iteration 13. [2026-03-25 20:54:56,797][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:54:56,798][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:55:08,983][__main__][INFO] - Number of regex retries in iteration 13: 0 [2026-03-25 20:55:08,984][__main__][INFO] - agents played in iteration 13 are Alice, Bob [2026-03-25 20:55:09,798][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:55:09,843][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:55:09,843][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:55:09,844][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:55:10,462][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:55:10,914][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:55:11,410][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:55:11,908][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:55:12,406][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:55:12,901][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:55:13,395][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:55:13,890][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:55:14,385][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:55:14,879][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:55:15,374][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:55:15,869][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:55:16,365][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:55:16,858][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:55:17,350][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:55:17,842][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:55:18,336][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:55:18,828][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:55:19,320][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:55:19,811][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:55:20,303][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:55:20,795][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:55:21,286][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:55:21,779][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:55:22,273][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:55:22,770][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:55:23,270][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:55:23,769][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:55:24,266][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:55:24,763][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:55:25,260][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:55:25,756][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:55:26,253][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:55:26,748][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:55:27,244][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:55:27,741][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:55:28,237][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:55:28,734][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:55:29,229][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:55:29,726][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:55:30,222][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:55:30,718][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:55:31,213][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:55:31,711][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:55:32,208][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:55:32,706][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:55:33,203][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:55:33,700][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:55:34,196][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:55:34,689][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:55:35,182][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:55:35,677][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:55:36,170][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:55:36,663][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:55:37,157][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:55:37,650][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:55:38,142][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:55:38,635][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:55:39,127][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:55:39,619][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:55:40,114][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:55:40,607][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:55:41,099][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:55:41,591][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:55:42,084][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:55:42,580][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:55:43,075][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:55:43,571][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:55:44,065][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:55:44,561][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:55:45,055][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:55:45,550][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:55:46,045][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:55:46,539][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:55:47,032][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:55:47,524][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:55:48,018][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:55:48,512][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:55:49,004][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:55:49,496][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:55:49,989][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:55:50,481][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:55:50,971][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:55:51,463][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:55:51,954][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:55:52,445][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:55:52,936][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:55:53,427][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:55:53,918][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:55:54,414][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:55:54,909][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:55:55,405][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:55:55,907][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:55:56,403][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:55:56,898][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:55:57,392][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:55:57,886][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:55:58,380][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:55:58,873][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:55:59,367][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:55:59,861][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:56:00,355][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:56:00,848][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:56:01,342][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:56:01,836][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:56:02,329][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:56:02,823][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:56:03,318][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:56:03,812][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:56:04,307][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:56:04,800][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:56:05,294][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:56:05,790][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:56:06,285][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:56:06,781][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:56:07,276][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:56:07,770][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:56:08,265][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:56:08,762][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:56:09,258][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:56:09,753][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:56:10,248][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:56:10,742][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:56:11,237][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:56:11,734][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:56:12,228][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:56:12,722][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:56:13,217][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:56:13,713][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19477 tokens. [2026-03-25 20:56:14,368][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.99%, Current % of VRAM taken: 60.35%, Block Peak % of device VRAM: 61.88%, ΔTime: 00:01:03 [2026-03-25 20:56:15,118][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:56:15,120][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:56:15,122][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:56:15,775][__main__][INFO] - Iteration 14 took 1m 18s (15.43% Gen, 83.74% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 27m 21s. Estimated total time: 65h 48m 55s. Time estimates for 10 more iterations: 13m 9s, 100 more iterations: 2h 11m 37s, 500 more iterations: 10h 58m 9s. [2026-03-25 20:56:15,778][__main__][INFO] - Starting iteration 14. [2026-03-25 20:56:16,188][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:56:16,189][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:56:16,862][mllm.models.large_language_model_local][WARNING] - Response Proposal: 5 hats, 5 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:56:20,991][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls Given Bob's high估值对帽子和书籍的价值以及你对书籍和球的低估值,你最好争取尽可能多的帽子,因为鲍勃更想得到帽子,而你更想得到书籍和球但这两者的相对价值对鲍勃来说较低。因此,所有帽子你可以争取最多,这样可以最大化你的得分。 did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 20:56:29,589][__main__][INFO] - Number of regex retries in iteration 14: 2 [2026-03-25 20:56:29,590][__main__][INFO] - agents played in iteration 14 are Alice, Bob [2026-03-25 20:56:30,410][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.38%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:56:30,455][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.38%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:56:30,456][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:56:30,456][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:56:31,073][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:56:31,522][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:56:32,024][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:56:32,519][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:56:33,012][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:56:33,506][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:56:34,000][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:56:34,492][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:56:34,985][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:56:35,477][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:56:35,969][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:56:36,461][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:56:36,953][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:56:37,445][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:56:37,938][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:56:38,431][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:56:38,924][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:56:39,417][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:56:39,909][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:56:40,403][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:56:40,896][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:56:41,390][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:56:41,884][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:56:42,381][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:56:42,878][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:56:43,370][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:56:43,862][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:56:44,354][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:56:44,847][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:56:45,340][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:56:45,834][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:56:46,326][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:56:46,818][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:56:47,312][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:56:47,806][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:56:48,302][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:56:48,795][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:56:49,289][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:56:49,785][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:56:50,281][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:56:50,777][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:56:51,272][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:56:51,766][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:56:52,260][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:56:52,754][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:56:53,248][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:56:53,744][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:56:54,237][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:56:54,734][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:56:55,229][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:56:55,725][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:56:56,221][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:56:56,717][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:56:57,211][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:56:57,707][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:56:58,202][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:56:58,696][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:56:59,192][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:56:59,687][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:57:00,183][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:57:00,677][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:57:01,169][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:57:01,662][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:57:02,154][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:57:02,646][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:57:03,139][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:57:03,632][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:57:04,124][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:57:04,616][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:57:05,109][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:57:05,602][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:57:06,094][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:57:06,588][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:57:07,081][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:57:07,574][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:57:08,065][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:57:08,559][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:57:09,051][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:57:09,543][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:57:10,035][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:57:10,527][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:57:11,019][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:57:11,513][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:57:12,007][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:57:12,500][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:57:12,994][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:57:13,487][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:57:13,982][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:57:14,477][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:57:14,969][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:57:15,462][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:57:15,954][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:57:16,445][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:57:16,937][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:57:17,429][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:57:17,921][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:57:18,415][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:57:18,910][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:57:19,404][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:57:19,899][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:57:20,392][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:57:20,885][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:57:21,379][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:57:21,872][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:57:22,364][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:57:22,857][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:57:23,350][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:57:23,842][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:57:24,333][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:57:24,825][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:57:25,318][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:57:25,810][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:57:26,304][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:57:26,797][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:57:27,290][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:57:27,784][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:57:28,279][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:57:28,771][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:57:29,265][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:57:29,758][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:57:30,250][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:57:30,742][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:57:31,234][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:57:31,727][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:57:32,220][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:57:32,711][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:57:33,207][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:57:33,704][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:57:34,200][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19402 tokens. [2026-03-25 20:57:34,858][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.95%, Current % of VRAM taken: 60.30%, Block Peak % of device VRAM: 61.78%, ΔTime: 00:01:03 [2026-03-25 20:57:35,638][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:57:35,641][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:57:35,642][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:57:36,401][__main__][INFO] - Iteration 15 took 1m 20s (16.71% Gen, 82.35% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 27m 47s. Estimated total time: 66h 50m 41s. Time estimates for 10 more iterations: 13m 22s, 100 more iterations: 2h 13m 41s, 500 more iterations: 11h 8m 26s. [2026-03-25 20:57:36,404][__main__][INFO] - Starting iteration 15. [2026-03-25 20:57:36,815][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:57:36,816][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:57:50,091][__main__][INFO] - Number of regex retries in iteration 15: 0 [2026-03-25 20:57:50,092][__main__][INFO] - agents played in iteration 15 are Alice, Bob [2026-03-25 20:57:50,907][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.35%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:57:50,951][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.35%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:57:50,952][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:57:50,953][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:57:51,557][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:57:52,008][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:57:52,507][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:57:53,002][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:57:53,499][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:57:53,994][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:57:54,489][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:57:54,983][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:57:55,478][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:57:55,972][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:57:56,467][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:57:56,962][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:57:57,459][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:57:57,952][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:57:58,446][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:57:58,941][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:57:59,434][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:57:59,927][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:58:00,422][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:58:00,914][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:58:01,407][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:58:01,901][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:58:02,394][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:58:02,886][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:58:03,381][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:58:03,873][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:58:04,368][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:58:04,862][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:58:05,354][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:58:05,846][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:58:06,340][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:58:06,835][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:58:07,328][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:58:07,819][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:58:08,311][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:58:08,803][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:58:09,295][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:58:09,788][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:58:10,281][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:58:10,773][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:58:11,265][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:58:11,759][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:58:12,252][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:58:12,746][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:58:13,241][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:58:13,734][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:58:14,226][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:58:14,719][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:58:15,214][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:58:15,710][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:58:16,204][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:58:16,699][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:58:17,192][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:58:17,686][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:58:18,182][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:58:18,675][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:58:19,169][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:58:19,663][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:58:20,157][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:58:20,650][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:58:21,144][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:58:21,640][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:58:22,135][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:58:22,630][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:58:23,127][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:58:23,623][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:58:24,119][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:58:24,615][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:58:25,111][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:58:25,607][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:58:26,103][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:58:26,601][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:58:27,097][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:58:27,594][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:58:28,090][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:58:28,586][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:58:29,083][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:58:29,579][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:58:30,074][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:58:30,569][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:58:31,065][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:58:31,560][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:58:32,056][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:58:32,551][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:58:33,047][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:58:33,541][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:58:34,034][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:58:34,528][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:58:35,020][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:58:35,513][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:58:36,008][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:58:36,502][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:58:36,994][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:58:37,487][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:58:37,980][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:58:38,473][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:58:38,967][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 20:58:39,459][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 20:58:39,950][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 20:58:40,442][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 20:58:40,934][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 20:58:41,426][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 20:58:41,920][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 20:58:42,412][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 20:58:42,905][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 20:58:43,399][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 20:58:43,891][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 20:58:44,385][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 20:58:44,879][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 20:58:45,372][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 20:58:45,865][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 20:58:46,358][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 20:58:46,852][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 20:58:47,346][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 20:58:47,840][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 20:58:48,335][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 20:58:48,829][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 20:58:49,323][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 20:58:49,816][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 20:58:50,310][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 20:58:50,804][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 20:58:51,296][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 20:58:51,790][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 20:58:52,284][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 20:58:52,780][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 20:58:53,273][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 20:58:53,766][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 20:58:54,259][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 20:58:54,752][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19427 tokens. [2026-03-25 20:58:55,383][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.00%, Current % of VRAM taken: 60.36%, Block Peak % of device VRAM: 61.71%, ΔTime: 00:01:03 [2026-03-25 20:58:56,144][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 20:58:56,147][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 20:58:56,148][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 20:58:56,841][__main__][INFO] - Iteration 16 took 1m 20s (16.59% Gen, 82.54% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 17m 3s. Estimated total time: 66h 41m 18s. Time estimates for 10 more iterations: 13m 20s, 100 more iterations: 2h 13m 22s, 500 more iterations: 11h 6m 53s. [2026-03-25 20:58:56,843][__main__][INFO] - Starting iteration 16. [2026-03-25 20:58:57,242][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 20:58:57,243][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 20:59:10,484][__main__][INFO] - Number of regex retries in iteration 16: 0 [2026-03-25 20:59:10,485][__main__][INFO] - agents played in iteration 16 are Alice, Bob [2026-03-25 20:59:11,523][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:59:11,568][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 20:59:11,569][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 20:59:11,569][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 20:59:12,192][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 20:59:12,643][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 20:59:13,144][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 20:59:13,641][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 20:59:14,136][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 20:59:14,632][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 20:59:15,127][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 20:59:15,625][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 20:59:16,124][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 20:59:16,624][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 20:59:17,124][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 20:59:17,624][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 20:59:18,124][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 20:59:18,622][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 20:59:19,120][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 20:59:19,615][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 20:59:20,130][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 20:59:20,625][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 20:59:21,119][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 20:59:21,613][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 20:59:22,107][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 20:59:22,603][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 20:59:23,102][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 20:59:23,600][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 20:59:24,094][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 20:59:24,586][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 20:59:25,080][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 20:59:25,572][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 20:59:26,064][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 20:59:26,555][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 20:59:27,047][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 20:59:27,537][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 20:59:28,029][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 20:59:28,523][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 20:59:29,016][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 20:59:29,510][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 20:59:30,005][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 20:59:30,499][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 20:59:30,992][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 20:59:31,488][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 20:59:31,983][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 20:59:32,477][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 20:59:32,969][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 20:59:33,462][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 20:59:33,955][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 20:59:34,448][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 20:59:34,940][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 20:59:35,433][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 20:59:35,925][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 20:59:36,418][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 20:59:36,910][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 20:59:37,402][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 20:59:37,893][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 20:59:38,384][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 20:59:38,875][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 20:59:39,366][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 20:59:39,857][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 20:59:40,351][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 20:59:40,847][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 20:59:41,341][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 20:59:41,835][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 20:59:42,329][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 20:59:42,824][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 20:59:43,316][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 20:59:43,809][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 20:59:44,302][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 20:59:44,795][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 20:59:45,288][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 20:59:45,783][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 20:59:46,275][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 20:59:46,767][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 20:59:47,260][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 20:59:47,754][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 20:59:48,249][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 20:59:48,743][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 20:59:49,237][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 20:59:49,730][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 20:59:50,223][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 20:59:50,716][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 20:59:51,209][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 20:59:51,706][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 20:59:52,204][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 20:59:52,702][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 20:59:53,198][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 20:59:53,694][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 20:59:54,191][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 20:59:54,688][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 20:59:55,185][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 20:59:55,681][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 20:59:56,176][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 20:59:56,670][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 20:59:57,165][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 20:59:57,661][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 20:59:58,156][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 20:59:58,650][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 20:59:59,145][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 20:59:59,641][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:00:00,136][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:00:00,633][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:00:01,129][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:00:01,624][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:00:02,121][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:00:02,617][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:00:03,111][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:00:03,606][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:00:04,100][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:00:04,592][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:00:05,087][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:00:05,580][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:00:06,072][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:00:06,564][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:00:07,058][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:00:07,550][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:00:08,043][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:00:08,534][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:00:09,025][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:00:09,517][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:00:10,007][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:00:10,500][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:00:10,991][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:00:11,484][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:00:11,977][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:00:12,472][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:00:12,967][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:00:13,461][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:00:13,955][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:00:14,448][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:00:14,941][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:00:15,433][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19467 tokens. [2026-03-25 21:00:16,071][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.94%, Current % of VRAM taken: 60.30%, Block Peak % of device VRAM: 61.88%, ΔTime: 00:01:03 [2026-03-25 21:00:16,829][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:00:16,832][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:00:16,833][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:00:17,719][__main__][INFO] - Iteration 17 took 1m 20s (16.45% Gen, 82.44% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 38m 15s. Estimated total time: 67h 3m 50s. Time estimates for 10 more iterations: 13m 24s, 100 more iterations: 2h 14m 7s, 500 more iterations: 11h 10m 38s. [2026-03-25 21:00:17,721][__main__][INFO] - Starting iteration 17. [2026-03-25 21:00:18,123][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:00:18,124][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:00:31,038][__main__][INFO] - Number of regex retries in iteration 17: 0 [2026-03-25 21:00:31,039][__main__][INFO] - agents played in iteration 17 are Alice, Bob [2026-03-25 21:00:31,855][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:00:31,899][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:00:31,900][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:00:31,900][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:00:32,510][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:00:32,962][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:00:33,465][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:00:33,962][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:00:34,458][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:00:34,960][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:00:35,457][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:00:35,955][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:00:36,453][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:00:36,949][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:00:37,445][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:00:37,941][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:00:38,436][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:00:38,931][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:00:39,427][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:00:39,923][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:00:40,420][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:00:40,915][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:00:41,410][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:00:41,905][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:00:42,401][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:00:42,897][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:00:43,392][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:00:43,888][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:00:44,384][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:00:44,882][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:00:45,379][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:00:45,877][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:00:46,379][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:00:46,874][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:00:47,369][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:00:47,865][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:00:48,363][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:00:48,860][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:00:49,354][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:00:49,850][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:00:50,346][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:00:50,842][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:00:51,335][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:00:51,829][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:00:52,325][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:00:52,818][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:00:53,312][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:00:53,806][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:00:54,303][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:00:54,798][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:00:55,292][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:00:55,788][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:00:56,282][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:00:56,775][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:00:57,267][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:00:57,759][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:00:58,252][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:00:58,744][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:00:59,236][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:00:59,728][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:01:00,220][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:01:00,712][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:01:01,205][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:01:01,699][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:01:02,193][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:01:02,687][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:01:03,183][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:01:03,678][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:01:04,170][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:01:04,665][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:01:05,166][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:01:05,662][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:01:06,159][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:01:06,656][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:01:07,150][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:01:07,646][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:01:08,142][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:01:08,635][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:01:09,128][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:01:09,621][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:01:10,113][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:01:10,606][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:01:11,102][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:01:11,598][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:01:12,092][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:01:12,587][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:01:13,080][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:01:13,574][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:01:14,066][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:01:14,560][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:01:15,052][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:01:15,549][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:01:16,045][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:01:16,539][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:01:17,033][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:01:17,527][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:01:18,026][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:01:18,523][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:01:19,021][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:01:19,518][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:01:20,016][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:01:20,518][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:01:21,014][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:01:21,511][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:01:22,008][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:01:22,505][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:01:23,000][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:01:23,497][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:01:23,997][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:01:24,492][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:01:24,988][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:01:25,485][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:01:25,982][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:01:26,479][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:01:26,973][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:01:27,472][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:01:27,968][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:01:28,464][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:01:28,960][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:01:29,454][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:01:29,948][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:01:30,440][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:01:30,932][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:01:31,423][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:01:31,915][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:01:32,409][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:01:32,902][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:01:33,396][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:01:33,889][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:01:34,384][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:01:34,877][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:01:35,371][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:01:35,865][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19596 tokens. [2026-03-25 21:01:36,505][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.00%, Current % of VRAM taken: 60.36%, Block Peak % of device VRAM: 61.92%, ΔTime: 00:01:04 [2026-03-25 21:01:37,256][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:01:37,259][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:01:37,260][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:01:37,970][__main__][INFO] - Iteration 18 took 1m 19s (16.17% Gen, 82.94% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 66h 5m 25s. Estimated total time: 66h 32m 21s. Time estimates for 10 more iterations: 13m 18s, 100 more iterations: 2h 13m 4s, 500 more iterations: 11h 5m 23s. [2026-03-25 21:01:37,972][__main__][INFO] - Starting iteration 18. [2026-03-25 21:01:38,374][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:01:38,375][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:01:51,013][__main__][INFO] - Number of regex retries in iteration 18: 0 [2026-03-25 21:01:51,014][__main__][INFO] - agents played in iteration 18 are Alice, Bob [2026-03-25 21:01:51,831][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:01:51,875][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:01:51,876][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:01:51,877][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:01:52,490][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:01:52,940][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:01:53,441][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:01:53,936][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:01:54,433][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:01:54,928][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:01:55,424][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:01:55,917][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:01:56,415][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:01:56,908][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:01:57,402][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:01:57,894][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:01:58,387][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:01:58,879][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:01:59,372][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:01:59,865][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:02:00,357][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:02:00,853][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:02:01,350][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:02:01,845][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:02:02,343][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:02:02,842][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:02:03,338][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:02:03,835][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:02:04,333][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:02:04,829][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:02:05,324][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:02:05,819][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:02:06,314][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:02:06,810][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:02:07,306][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:02:07,804][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:02:08,300][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:02:08,797][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:02:09,291][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:02:09,788][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:02:10,285][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:02:10,780][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:02:11,277][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:02:11,772][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:02:12,269][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:02:12,766][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:02:13,264][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:02:13,758][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:02:14,251][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:02:14,744][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:02:15,238][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:02:15,731][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:02:16,224][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:02:16,718][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:02:17,211][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:02:17,705][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:02:18,200][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:02:18,695][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:02:19,189][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:02:19,685][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:02:20,180][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:02:20,673][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:02:21,168][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:02:21,664][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:02:22,159][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:02:22,652][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:02:23,147][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:02:23,641][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:02:24,135][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:02:24,631][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:02:25,127][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:02:25,623][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:02:26,117][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:02:26,611][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:02:27,107][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:02:27,601][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:02:28,097][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:02:28,590][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:02:29,084][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:02:29,578][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:02:30,070][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:02:30,563][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:02:31,056][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:02:31,548][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:02:32,041][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:02:32,537][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:02:33,034][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:02:33,529][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:02:34,021][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:02:34,517][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:02:35,011][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:02:35,506][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:02:36,001][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:02:36,494][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:02:36,986][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:02:37,484][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:02:37,978][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:02:38,470][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:02:38,965][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:02:39,460][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:02:39,953][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:02:40,448][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:02:40,939][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:02:41,434][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:02:41,927][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:02:42,421][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:02:42,915][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:02:43,408][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:02:43,901][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:02:44,393][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:02:44,886][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:02:45,379][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:02:45,873][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:02:46,369][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:02:46,863][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:02:47,356][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:02:47,851][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:02:48,348][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:02:48,843][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:02:49,338][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:02:49,833][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:02:50,329][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:02:50,824][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:02:51,320][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:02:51,816][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:02:52,312][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:02:52,809][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:02:53,305][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:02:53,802][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:02:54,298][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:02:54,795][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:02:55,291][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:02:55,789][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19631 tokens. [2026-03-25 21:02:56,437][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.08%, Current % of VRAM taken: 60.43%, Block Peak % of device VRAM: 61.95%, ΔTime: 00:01:03 [2026-03-25 21:02:57,180][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:02:57,182][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:02:57,184][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:02:57,826][__main__][INFO] - Iteration 19 took 1m 19s (15.91% Gen, 83.28% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 44m 22s. Estimated total time: 66h 12m 38s. Time estimates for 10 more iterations: 13m 14s, 100 more iterations: 2h 12m 25s, 500 more iterations: 11h 2m 6s. [2026-03-25 21:02:57,829][__main__][INFO] - Starting iteration 19. [2026-03-25 21:02:58,227][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:02:58,228][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:03:10,325][__main__][INFO] - Number of regex retries in iteration 19: 0 [2026-03-25 21:03:10,326][__main__][INFO] - agents played in iteration 19 are Alice, Bob [2026-03-25 21:03:11,148][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.42%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:03:11,192][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.42%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:03:11,193][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:03:11,194][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:03:11,807][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:03:12,259][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:03:12,754][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:03:13,249][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:03:13,744][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:03:14,239][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:03:14,733][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:03:15,227][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:03:15,723][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:03:16,221][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:03:16,716][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:03:17,211][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:03:17,704][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:03:18,200][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:03:18,696][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:03:19,191][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:03:19,687][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:03:20,183][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:03:20,678][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:03:21,173][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:03:21,667][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:03:22,162][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:03:22,657][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:03:23,151][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:03:23,645][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:03:24,138][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:03:24,629][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:03:25,121][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:03:25,613][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:03:26,105][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:03:26,596][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:03:27,087][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:03:27,579][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:03:28,074][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:03:28,570][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:03:29,072][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:03:29,571][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:03:30,069][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:03:30,567][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:03:31,064][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:03:31,560][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:03:32,057][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:03:32,551][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:03:33,047][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:03:33,541][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:03:34,036][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:03:34,531][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:03:35,027][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:03:35,524][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:03:36,020][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:03:36,518][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:03:37,016][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:03:37,512][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:03:38,008][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:03:38,504][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:03:39,001][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:03:39,502][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:03:39,999][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:03:40,495][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:03:40,991][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:03:41,484][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:03:41,979][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:03:42,472][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:03:42,964][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:03:43,456][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:03:43,951][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:03:44,443][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:03:44,937][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:03:45,431][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:03:45,923][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:03:46,417][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:03:46,911][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:03:47,406][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:03:47,905][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:03:48,402][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:03:48,897][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:03:49,392][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:03:49,886][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:03:50,382][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:03:50,876][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:03:51,372][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:03:51,867][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:03:52,362][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:03:52,856][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:03:53,350][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:03:53,846][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:03:54,341][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:03:54,836][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:03:55,331][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:03:55,824][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:03:56,316][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:03:56,809][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:03:57,302][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:03:57,794][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:03:58,287][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:03:58,779][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:03:59,271][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:03:59,766][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:04:00,261][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:04:00,757][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:04:01,252][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:04:01,748][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:04:02,244][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:04:02,740][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:04:03,235][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:04:03,729][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:04:04,223][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:04:04,716][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:04:05,208][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:04:05,703][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:04:06,197][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:04:06,691][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:04:07,186][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:04:07,679][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:04:08,175][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:04:08,671][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:04:09,166][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:04:09,658][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:04:10,153][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:04:10,648][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:04:11,147][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:04:11,642][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:04:12,135][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:04:12,629][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:04:13,125][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:04:13,620][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:04:14,115][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:04:14,609][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:04:15,104][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19516 tokens. [2026-03-25 21:04:15,766][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.98%, Current % of VRAM taken: 60.34%, Block Peak % of device VRAM: 61.94%, ΔTime: 00:01:03 [2026-03-25 21:04:16,832][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:04:16,834][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:04:16,836][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:04:17,494][__main__][INFO] - Iteration 20 took 1m 19s (15.26% Gen, 83.91% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 33m 47s. Estimated total time: 66h 3m 22s. Time estimates for 10 more iterations: 13m 12s, 100 more iterations: 2h 12m 6s, 500 more iterations: 11h 0m 33s. [2026-03-25 21:04:17,497][__main__][INFO] - Starting iteration 20. [2026-03-25 21:04:17,911][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:04:17,912][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:04:25,573][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls Given Alice's per-item values (hats=10, books=1, balls=10), it seems she values hats and balls the most, and books the least. Since we value hats the most at 1, and both books and balls are valued at 10 by Alice, we should aim to keep as many hats as possible to maximize our score. This proposal ensures we get all 10 hats, which should yield the highest individual value given our per-item values. did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:04:27,209][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:04:32,203][__main__][INFO] - Number of regex retries in iteration 20: 2 [2026-03-25 21:04:32,205][__main__][INFO] - agents played in iteration 20 are Alice, Bob [2026-03-25 21:04:33,018][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.38%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:04:33,063][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.38%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:04:33,064][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:04:33,064][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:04:33,688][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:04:34,141][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:04:34,637][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:04:35,132][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:04:35,628][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:04:36,124][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:04:36,618][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:04:37,112][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:04:37,608][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:04:38,102][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:04:38,595][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:04:39,089][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:04:39,583][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:04:40,076][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:04:40,569][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:04:41,064][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:04:41,557][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:04:42,052][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:04:42,548][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:04:43,043][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:04:43,538][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:04:44,032][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:04:44,528][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:04:45,024][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:04:45,522][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:04:46,018][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:04:46,514][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:04:47,010][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:04:47,506][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:04:48,001][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:04:48,497][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:04:48,991][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:04:49,485][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:04:49,979][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:04:50,473][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:04:50,968][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:04:51,464][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:04:51,959][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:04:52,454][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:04:52,948][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:04:53,443][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:04:53,937][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:04:54,430][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:04:54,925][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:04:55,419][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:04:55,914][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:04:56,408][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:04:56,903][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:04:57,397][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:04:57,890][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:04:58,384][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:04:58,875][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:04:59,368][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:04:59,862][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:05:00,354][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:05:00,846][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:05:01,339][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:05:01,836][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:05:02,334][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:05:02,830][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:05:03,325][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:05:03,822][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:05:04,317][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:05:04,812][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:05:05,309][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:05:05,806][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:05:06,304][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:05:06,801][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:05:07,298][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:05:07,794][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:05:08,291][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:05:08,787][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:05:09,286][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:05:09,782][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:05:10,276][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:05:10,772][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:05:11,268][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:05:11,766][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:05:12,263][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:05:12,759][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:05:13,255][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:05:13,749][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:05:14,244][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:05:14,739][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:05:15,235][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:05:15,730][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:05:16,232][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:05:16,727][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:05:17,223][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:05:17,720][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:05:18,217][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:05:18,712][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:05:19,208][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:05:19,704][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:05:20,198][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:05:20,694][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:05:21,187][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:05:21,681][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:05:22,175][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:05:22,670][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:05:23,166][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:05:23,661][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:05:24,157][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:05:24,652][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:05:25,148][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:05:25,643][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:05:26,136][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:05:26,630][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:05:27,125][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:05:27,619][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:05:28,114][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:05:28,608][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:05:29,102][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:05:29,595][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:05:30,087][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:05:30,580][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:05:31,073][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:05:31,566][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:05:32,059][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:05:32,551][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:05:33,043][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:05:33,537][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:05:34,031][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:05:34,525][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:05:35,018][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:05:35,512][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:05:36,006][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:05:36,500][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:05:36,994][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19580 tokens. [2026-03-25 21:05:37,648][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.96%, Current % of VRAM taken: 60.31%, Block Peak % of device VRAM: 61.90%, ΔTime: 00:01:03 [2026-03-25 21:05:38,399][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:05:38,401][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:05:38,403][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:05:39,058][__main__][INFO] - Iteration 21 took 1m 21s (17.61% Gen, 81.58% Train). Generation: 14s, Training: 1m 6s. Estimated remaining time: 67h 6m 26s. Estimated total time: 67h 37m 23s. Time estimates for 10 more iterations: 13m 31s, 100 more iterations: 2h 15m 14s, 500 more iterations: 11h 16m 13s. [2026-03-25 21:05:39,062][__main__][INFO] - Starting iteration 21. [2026-03-25 21:05:39,465][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:05:39,466][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:05:53,190][__main__][INFO] - Number of regex retries in iteration 21: 0 [2026-03-25 21:05:53,191][__main__][INFO] - agents played in iteration 21 are Alice, Bob [2026-03-25 21:05:54,221][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:05:54,267][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.36%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:05:54,268][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:05:54,268][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:05:54,878][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:05:55,331][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:05:55,832][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:05:56,329][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:05:56,827][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:05:57,324][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:05:57,823][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:05:58,320][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:05:58,817][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:05:59,315][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:05:59,812][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:06:00,309][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:06:00,805][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:06:01,300][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:06:01,794][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:06:02,288][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:06:02,784][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:06:03,277][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:06:03,771][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:06:04,266][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:06:04,759][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:06:05,253][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:06:05,747][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:06:06,241][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:06:06,735][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:06:07,228][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:06:07,722][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:06:08,213][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:06:08,708][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:06:09,202][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:06:09,695][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:06:10,189][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:06:10,682][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:06:11,175][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:06:11,668][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:06:12,162][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:06:12,655][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:06:13,148][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:06:13,641][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:06:14,133][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:06:14,625][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:06:15,118][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:06:15,613][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:06:16,108][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:06:16,603][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:06:17,098][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:06:17,596][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:06:18,090][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:06:18,584][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:06:19,081][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:06:19,576][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:06:20,071][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:06:20,567][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:06:21,065][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:06:21,563][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:06:22,058][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:06:22,553][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:06:23,046][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:06:23,539][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:06:24,031][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:06:24,524][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:06:25,018][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:06:25,511][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:06:26,004][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:06:26,497][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:06:26,991][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:06:27,487][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:06:27,983][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:06:28,477][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:06:28,972][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:06:29,468][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:06:29,964][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:06:30,459][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:06:30,952][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:06:31,445][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:06:31,938][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:06:32,431][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:06:32,927][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:06:33,424][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:06:33,919][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:06:34,415][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:06:34,910][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:06:35,406][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:06:35,902][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:06:36,397][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:06:36,893][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:06:37,388][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:06:37,884][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:06:38,379][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:06:38,873][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:06:39,367][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:06:39,860][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:06:40,355][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:06:40,849][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:06:41,345][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:06:41,838][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:06:42,331][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:06:42,827][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:06:43,322][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:06:43,817][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:06:44,311][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:06:44,804][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:06:45,297][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:06:45,789][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:06:46,280][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:06:46,774][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:06:47,268][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:06:47,762][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:06:48,256][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:06:48,750][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:06:49,246][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:06:49,740][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:06:50,234][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:06:50,730][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:06:51,224][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:06:51,719][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:06:52,214][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:06:52,708][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:06:53,225][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:06:53,720][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:06:54,215][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:06:54,708][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:06:55,200][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:06:55,692][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:06:56,185][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:06:56,678][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:06:57,169][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:06:57,662][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:06:58,154][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19532 tokens. [2026-03-25 21:06:58,781][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.02%, Current % of VRAM taken: 60.38%, Block Peak % of device VRAM: 61.89%, ΔTime: 00:01:03 [2026-03-25 21:06:59,519][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:06:59,521][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:06:59,523][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:07:00,167][__main__][INFO] - Iteration 22 took 1m 20s (17.01% Gen, 82.19% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 42m 48s. Estimated total time: 67h 15m 6s. Time estimates for 10 more iterations: 13m 27s, 100 more iterations: 2h 14m 30s, 500 more iterations: 11h 12m 31s. [2026-03-25 21:07:00,169][__main__][INFO] - Starting iteration 22. [2026-03-25 21:07:00,570][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:07:00,571][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:07:08,815][mllm.models.large_language_model_local][WARNING] - Response Proposal: 5 hats, 0 books, 15 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:07:13,583][__main__][INFO] - Number of regex retries in iteration 22: 1 [2026-03-25 21:07:13,583][__main__][INFO] - agents played in iteration 22 are Alice, Bob [2026-03-25 21:07:14,413][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:07:14,458][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:07:14,459][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:07:14,460][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:07:15,072][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:07:15,525][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:07:16,031][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:07:16,527][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:07:17,025][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:07:17,520][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:07:18,014][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:07:18,511][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:07:19,008][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:07:19,506][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:07:20,003][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:07:20,506][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:07:21,005][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:07:21,503][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:07:21,999][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:07:22,495][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:07:22,991][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:07:23,486][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:07:23,981][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:07:24,476][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:07:24,971][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:07:25,465][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:07:25,959][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:07:26,455][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:07:26,951][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:07:27,447][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:07:27,942][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:07:28,440][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:07:28,934][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:07:29,431][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:07:29,924][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:07:30,419][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:07:30,914][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:07:31,408][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:07:31,903][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:07:32,399][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:07:32,893][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:07:33,388][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:07:33,883][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:07:34,377][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:07:34,874][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:07:35,370][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:07:35,864][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:07:36,357][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:07:36,852][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:07:37,348][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:07:37,844][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:07:38,338][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:07:38,833][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:07:39,327][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:07:39,820][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:07:40,313][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:07:40,806][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:07:41,298][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:07:41,791][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:07:42,285][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:07:42,777][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:07:43,272][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:07:43,765][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:07:44,258][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:07:44,750][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:07:45,268][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:07:45,761][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:07:46,254][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:07:46,751][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:07:47,245][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:07:47,738][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:07:48,233][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:07:48,728][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:07:49,222][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:07:49,714][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:07:50,208][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:07:50,703][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:07:51,198][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:07:51,693][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:07:52,195][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:07:52,690][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:07:53,185][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:07:53,678][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:07:54,173][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:07:54,667][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:07:55,159][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:07:55,651][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:07:56,146][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:07:56,637][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:07:57,131][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:07:57,624][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:07:58,116][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:07:58,609][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:07:59,102][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:07:59,597][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:08:00,094][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:08:00,592][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:08:01,091][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:08:01,587][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:08:02,085][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:08:02,582][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:08:03,079][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:08:03,574][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:08:04,071][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:08:04,569][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:08:05,066][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:08:05,563][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:08:06,059][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:08:06,557][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:08:07,056][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:08:07,552][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:08:08,048][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:08:08,545][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:08:09,042][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:08:09,540][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:08:10,035][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:08:10,534][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:08:11,031][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:08:11,528][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:08:12,024][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:08:12,516][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:08:13,009][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:08:13,504][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:08:13,997][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:08:14,491][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:08:14,985][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:08:15,478][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:08:15,973][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:08:16,465][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:08:16,959][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:08:17,452][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:08:17,948][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:08:18,445][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19776 tokens. [2026-03-25 21:08:19,086][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.12%, Current % of VRAM taken: 60.48%, Block Peak % of device VRAM: 61.97%, ΔTime: 00:01:04 [2026-03-25 21:08:19,830][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:08:19,833][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:08:19,834][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:08:20,540][__main__][INFO] - Iteration 23 took 1m 19s (16.27% Gen, 82.84% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 4m 52s. Estimated total time: 66h 38m 30s. Time estimates for 10 more iterations: 13m 19s, 100 more iterations: 2h 13m 17s, 500 more iterations: 11h 6m 25s. [2026-03-25 21:08:20,542][__main__][INFO] - Starting iteration 23. [2026-03-25 21:08:20,944][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:08:20,945][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:08:33,509][__main__][INFO] - Number of regex retries in iteration 23: 0 [2026-03-25 21:08:33,510][__main__][INFO] - agents played in iteration 23 are Alice, Bob [2026-03-25 21:08:34,334][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:08:34,379][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:08:34,380][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:08:34,380][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:08:35,001][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:08:35,453][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:08:35,951][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:08:36,444][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:08:36,940][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:08:37,433][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:08:37,928][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:08:38,422][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:08:38,915][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:08:39,409][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:08:39,903][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:08:40,397][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:08:40,890][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:08:41,383][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:08:41,879][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:08:42,372][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:08:42,872][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:08:43,366][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:08:43,859][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:08:44,354][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:08:44,850][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:08:45,345][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:08:45,843][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:08:46,338][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:08:46,832][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:08:47,328][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:08:47,828][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:08:48,325][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:08:48,821][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:08:49,318][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:08:49,818][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:08:50,314][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:08:50,810][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:08:51,306][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:08:51,803][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:08:52,300][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:08:52,796][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:08:53,291][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:08:53,788][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:08:54,286][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:08:54,783][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:08:55,281][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:08:55,776][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:08:56,271][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:08:56,764][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:08:57,258][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:08:57,752][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:08:58,248][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:08:58,741][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:08:59,234][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:08:59,728][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:09:00,221][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:09:00,715][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:09:01,209][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:09:01,703][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:09:02,197][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:09:02,690][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:09:03,185][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:09:03,678][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:09:04,171][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:09:04,664][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:09:05,158][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:09:05,651][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:09:06,144][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:09:06,642][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:09:07,136][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:09:07,630][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:09:08,122][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:09:08,617][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:09:09,110][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:09:09,604][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:09:10,096][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:09:10,589][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:09:11,084][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:09:11,576][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:09:12,071][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:09:12,565][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:09:13,059][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:09:13,552][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:09:14,046][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:09:14,541][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:09:15,035][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:09:15,527][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:09:16,019][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:09:16,513][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:09:17,006][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:09:17,498][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:09:17,991][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:09:18,487][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:09:18,981][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:09:19,474][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:09:19,969][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:09:20,465][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:09:20,959][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:09:21,455][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:09:21,951][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:09:22,446][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:09:22,939][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:09:23,433][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:09:23,926][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:09:24,420][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:09:24,913][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:09:25,408][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:09:25,902][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:09:26,397][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:09:26,890][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:09:27,385][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:09:27,878][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:09:28,375][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:09:28,872][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:09:29,368][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:09:29,866][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:09:30,362][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:09:30,857][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:09:31,352][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:09:31,846][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:09:32,341][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:09:32,837][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:09:33,332][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:09:33,827][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:09:34,321][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:09:34,816][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:09:35,311][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:09:35,806][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:09:36,302][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:09:36,799][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:09:37,295][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:09:37,789][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:09:38,285][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19614 tokens. [2026-03-25 21:09:38,946][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.03%, Current % of VRAM taken: 60.38%, Block Peak % of device VRAM: 61.83%, ΔTime: 00:01:03 [2026-03-25 21:09:39,695][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:09:39,697][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:09:39,699][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:09:40,572][__main__][INFO] - Iteration 24 took 1m 19s (15.78% Gen, 83.12% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 46m 26s. Estimated total time: 66h 21m 24s. Time estimates for 10 more iterations: 13m 16s, 100 more iterations: 2h 12m 42s, 500 more iterations: 11h 3m 34s. [2026-03-25 21:09:40,574][__main__][INFO] - Starting iteration 24. [2026-03-25 21:09:40,979][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:09:40,980][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:09:53,766][__main__][INFO] - Number of regex retries in iteration 24: 0 [2026-03-25 21:09:53,766][__main__][INFO] - agents played in iteration 24 are Alice, Bob [2026-03-25 21:09:54,590][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:09:54,635][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.37%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:09:54,636][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:09:54,636][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:09:55,253][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:09:55,704][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:09:56,205][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:09:56,699][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:09:57,192][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:09:57,689][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:09:58,182][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:09:58,675][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:09:59,169][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:09:59,667][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:10:00,181][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:10:00,676][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:10:01,172][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:10:01,667][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:10:02,161][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:10:02,656][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:10:03,150][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:10:03,642][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:10:04,135][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:10:04,629][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:10:05,125][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:10:05,619][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:10:06,113][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:10:06,611][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:10:07,106][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:10:07,598][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:10:08,092][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:10:08,584][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:10:09,076][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:10:09,568][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:10:10,060][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:10:10,553][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:10:11,044][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:10:11,539][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:10:12,033][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:10:12,530][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:10:13,028][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:10:13,523][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:10:14,020][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:10:14,516][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:10:15,014][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:10:15,511][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:10:16,008][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:10:16,505][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:10:17,000][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:10:17,495][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:10:17,992][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:10:18,491][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:10:18,988][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:10:19,486][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:10:19,983][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:10:20,480][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:10:20,976][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:10:21,473][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:10:21,969][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:10:22,466][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:10:22,963][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:10:23,458][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:10:23,953][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:10:24,447][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:10:24,942][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:10:25,435][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:10:25,930][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:10:26,426][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:10:26,919][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:10:27,414][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:10:27,908][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:10:28,402][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:10:28,896][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:10:29,390][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:10:29,883][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:10:30,376][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:10:30,870][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:10:31,366][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:10:31,861][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:10:32,355][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:10:32,849][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:10:33,345][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:10:33,840][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:10:34,334][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:10:34,828][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:10:35,322][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:10:35,815][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:10:36,309][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:10:36,803][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:10:37,296][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:10:37,790][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:10:38,286][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:10:38,780][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:10:39,271][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:10:39,763][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:10:40,255][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:10:40,748][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:10:41,239][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:10:41,730][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:10:42,222][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:10:42,713][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:10:43,208][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:10:43,705][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:10:44,201][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:10:44,696][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:10:45,191][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:10:45,686][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:10:46,181][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:10:46,675][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:10:47,170][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:10:47,663][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:10:48,157][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:10:48,653][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:10:49,146][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:10:49,639][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:10:50,132][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:10:50,625][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:10:51,119][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:10:51,614][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:10:52,107][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:10:52,601][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:10:53,096][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:10:53,590][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:10:54,085][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:10:54,579][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:10:55,073][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:10:55,568][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:10:56,062][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:10:56,559][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:10:57,054][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:10:57,551][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:10:58,049][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:10:58,546][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19795 tokens. [2026-03-25 21:10:59,197][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.01%, Current % of VRAM taken: 60.37%, Block Peak % of device VRAM: 61.81%, ΔTime: 00:01:03 [2026-03-25 21:10:59,956][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:10:59,959][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:10:59,960][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:11:00,824][__main__][INFO] - Iteration 25 took 1m 19s (16.01% Gen, 82.90% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 55m 56s. Estimated total time: 66h 32m 15s. Time estimates for 10 more iterations: 13m 18s, 100 more iterations: 2h 13m 4s, 500 more iterations: 11h 5m 22s. [2026-03-25 21:11:00,828][__main__][INFO] - Starting iteration 25. [2026-03-25 21:11:01,244][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:11:01,244][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:11:02,014][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:11:06,875][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls Given the values, I (Alice) have a much higher value for hats compared to Bob, while Bob values books and balls more. To maximize my points, it makes sense to claim all the hats since I have a significant advantage in valuing them over the other items. For books and balls, I would offer nothing and let Bob take them, as he values them more. did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:11:15,606][__main__][INFO] - Number of regex retries in iteration 25: 2 [2026-03-25 21:11:15,607][__main__][INFO] - agents played in iteration 25 are Alice, Bob [2026-03-25 21:11:16,417][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:11:16,462][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:11:16,463][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:11:16,463][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:11:17,081][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:11:17,530][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:11:18,033][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:11:18,533][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:11:19,028][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:11:19,523][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:11:20,017][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:11:20,512][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:11:21,008][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:11:21,502][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:11:21,997][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:11:22,492][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:11:22,986][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:11:23,479][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:11:23,973][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:11:24,467][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:11:24,961][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:11:25,455][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:11:25,948][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:11:26,441][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:11:26,933][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:11:27,425][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:11:27,918][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:11:28,410][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:11:28,903][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:11:29,400][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:11:29,894][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:11:30,389][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:11:30,887][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:11:31,380][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:11:31,874][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:11:32,369][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:11:32,865][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:11:33,358][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:11:33,852][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:11:34,347][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:11:34,841][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:11:35,334][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:11:35,829][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:11:36,323][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:11:36,817][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:11:37,311][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:11:37,806][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:11:38,300][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:11:38,795][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:11:39,291][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:11:39,786][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:11:40,280][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:11:40,775][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:11:41,271][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:11:41,766][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:11:42,266][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:11:42,760][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:11:43,255][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:11:43,756][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:11:44,259][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:11:44,756][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:11:45,251][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:11:45,746][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:11:46,242][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:11:46,742][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:11:47,238][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:11:47,735][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:11:48,231][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:11:48,729][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:11:49,226][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:11:49,723][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:11:50,223][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:11:50,720][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:11:51,217][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:11:51,714][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:11:52,211][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:11:52,709][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:11:53,205][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:11:53,701][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:11:54,196][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:11:54,692][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:11:55,188][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:11:55,682][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:11:56,178][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:11:56,675][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:11:57,172][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:11:57,666][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:11:58,159][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:11:58,651][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:11:59,143][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:11:59,636][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:12:00,130][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:12:00,622][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:12:01,115][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:12:01,610][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:12:02,106][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:12:02,600][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:12:03,095][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:12:03,590][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:12:04,085][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:12:04,578][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:12:05,072][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:12:05,566][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:12:06,064][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:12:06,558][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:12:07,052][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:12:07,547][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:12:08,040][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:12:08,533][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:12:09,028][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:12:09,524][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:12:10,019][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:12:10,515][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:12:11,009][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:12:11,505][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:12:12,001][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:12:12,496][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:12:12,991][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:12:13,486][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:12:13,982][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:12:14,475][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:12:14,971][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:12:15,467][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:12:15,964][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:12:16,459][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:12:16,954][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:12:17,449][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:12:17,943][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:12:18,439][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:12:18,934][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:12:19,428][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:12:19,922][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:12:20,416][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19846 tokens. [2026-03-25 21:12:21,073][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.04%, Current % of VRAM taken: 60.40%, Block Peak % of device VRAM: 62.03%, ΔTime: 00:01:04 [2026-03-25 21:12:21,870][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:12:21,872][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:12:21,874][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:12:22,738][__main__][INFO] - Iteration 26 took 1m 21s (17.62% Gen, 81.31% Train). Generation: 14s, Training: 1m 6s. Estimated remaining time: 67h 17m 3s. Estimated total time: 67h 54m 44s. Time estimates for 10 more iterations: 13m 34s, 100 more iterations: 2h 15m 49s, 500 more iterations: 11h 19m 7s. [2026-03-25 21:12:22,740][__main__][INFO] - Starting iteration 26. [2026-03-25 21:12:23,141][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:12:23,141][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:12:36,099][__main__][INFO] - Number of regex retries in iteration 26: 0 [2026-03-25 21:12:36,099][__main__][INFO] - agents played in iteration 26 are Alice, Bob [2026-03-25 21:12:37,132][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:12:37,177][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:12:37,178][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:12:37,178][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:12:37,793][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:12:38,244][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:12:38,746][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:12:39,242][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:12:39,737][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:12:40,232][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:12:40,727][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:12:41,222][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:12:41,718][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:12:42,212][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:12:42,707][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:12:43,204][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:12:43,701][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:12:44,197][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:12:44,694][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:12:45,188][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:12:45,680][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:12:46,176][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:12:46,672][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:12:47,167][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:12:47,661][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:12:48,157][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:12:48,652][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:12:49,148][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:12:49,642][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:12:50,137][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:12:50,632][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:12:51,127][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:12:51,622][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:12:52,117][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:12:52,612][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:12:53,105][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:12:53,598][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:12:54,092][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:12:54,588][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:12:55,083][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:12:55,576][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:12:56,071][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:12:56,567][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:12:57,062][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:12:57,555][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:12:58,049][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:12:58,540][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:12:59,032][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:12:59,525][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:13:00,017][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:13:00,510][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:13:01,002][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:13:01,494][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:13:01,988][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:13:02,481][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:13:02,974][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:13:03,467][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:13:03,960][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:13:04,454][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:13:04,948][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:13:05,440][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:13:05,934][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:13:06,429][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:13:06,923][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:13:07,416][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:13:07,910][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:13:08,405][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:13:08,898][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:13:09,393][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:13:09,886][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:13:10,380][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:13:10,874][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:13:11,368][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:13:11,859][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:13:12,352][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:13:12,844][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:13:13,338][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:13:13,830][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:13:14,322][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:13:14,816][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:13:15,310][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:13:15,804][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:13:16,297][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:13:16,790][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:13:17,288][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:13:17,787][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:13:18,285][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:13:18,781][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:13:19,277][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:13:19,773][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:13:20,269][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:13:20,764][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:13:21,260][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:13:21,756][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:13:22,254][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:13:22,751][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:13:23,248][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:13:23,745][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:13:24,242][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:13:24,736][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:13:25,233][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:13:25,729][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:13:26,226][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:13:26,722][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:13:27,217][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:13:27,715][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:13:28,235][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:13:28,736][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:13:29,231][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:13:29,722][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:13:30,214][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:13:30,706][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:13:31,197][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:13:31,691][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:13:32,182][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:13:32,675][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:13:33,167][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:13:33,660][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:13:34,155][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:13:34,649][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:13:35,141][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:13:35,634][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:13:36,128][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:13:36,620][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:13:37,113][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:13:37,608][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:13:38,102][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:13:38,596][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:13:39,090][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:13:39,584][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:13:40,078][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:13:40,573][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:13:41,069][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19790 tokens. [2026-03-25 21:13:41,700][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.08%, Current % of VRAM taken: 60.44%, Block Peak % of device VRAM: 61.91%, ΔTime: 00:01:03 [2026-03-25 21:13:42,442][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:13:42,444][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:13:42,445][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:13:43,106][__main__][INFO] - Iteration 27 took 1m 19s (16.20% Gen, 82.97% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 59m 17s. Estimated total time: 66h 38m 18s. Time estimates for 10 more iterations: 13m 19s, 100 more iterations: 2h 13m 16s, 500 more iterations: 11h 6m 23s. [2026-03-25 21:13:43,109][__main__][INFO] - Starting iteration 27. [2026-03-25 21:13:43,506][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:13:43,507][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:13:56,599][__main__][INFO] - Number of regex retries in iteration 27: 0 [2026-03-25 21:13:56,600][__main__][INFO] - agents played in iteration 27 are Alice, Bob [2026-03-25 21:13:57,423][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:13:57,468][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:13:57,468][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:13:57,469][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:13:58,086][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:13:58,537][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:13:59,037][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:13:59,532][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:14:00,026][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:14:00,521][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:14:01,015][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:14:01,512][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:14:02,011][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:14:02,508][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:14:03,003][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:14:03,500][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:14:03,994][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:14:04,489][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:14:04,986][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:14:05,481][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:14:05,977][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:14:06,472][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:14:06,969][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:14:07,466][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:14:07,962][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:14:08,457][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:14:08,953][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:14:09,450][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:14:09,945][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:14:10,441][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:14:10,937][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:14:11,435][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:14:11,933][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:14:12,431][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:14:12,929][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:14:13,426][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:14:13,922][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:14:14,419][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:14:14,912][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:14:15,410][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:14:15,903][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:14:16,395][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:14:16,889][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:14:17,383][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:14:17,878][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:14:18,375][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:14:18,871][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:14:19,367][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:14:19,866][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:14:20,365][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:14:20,859][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:14:21,356][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:14:21,851][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:14:22,343][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:14:22,835][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:14:23,329][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:14:23,822][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:14:24,315][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:14:24,807][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:14:25,300][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:14:25,792][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:14:26,288][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:14:26,782][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:14:27,277][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:14:27,770][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:14:28,262][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:14:28,757][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:14:29,251][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:14:29,745][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:14:30,240][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:14:30,736][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:14:31,230][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:14:31,725][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:14:32,220][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:14:32,715][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:14:33,210][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:14:33,704][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:14:34,197][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:14:34,691][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:14:35,187][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:14:35,679][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:14:36,172][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:14:36,668][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:14:37,161][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:14:37,655][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:14:38,149][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:14:38,643][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:14:39,140][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:14:39,634][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:14:40,126][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:14:40,619][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:14:41,114][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:14:41,606][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:14:42,101][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:14:42,596][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:14:43,092][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:14:43,590][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:14:44,084][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:14:44,579][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:14:45,076][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:14:45,577][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:14:46,073][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:14:46,568][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:14:47,064][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:14:47,560][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:14:48,053][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:14:48,549][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:14:49,044][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:14:49,540][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:14:50,037][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:14:50,534][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:14:51,031][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:14:51,530][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:14:52,026][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:14:52,523][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:14:53,020][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:14:53,516][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:14:54,014][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:14:54,509][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:14:55,007][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:14:55,502][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:14:55,999][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:14:56,496][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:14:56,990][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:14:57,482][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:14:57,976][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:14:58,470][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:14:58,963][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:14:59,458][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:14:59,951][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:15:00,446][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:15:00,939][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:15:01,431][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19826 tokens. [2026-03-25 21:15:02,071][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 13.99%, Current % of VRAM taken: 60.35%, Block Peak % of device VRAM: 61.97%, ΔTime: 00:01:03 [2026-03-25 21:15:02,808][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:15:02,810][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:15:02,812][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:15:03,517][__main__][INFO] - Iteration 28 took 1m 20s (16.36% Gen, 82.75% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 0m 11s. Estimated total time: 66h 40m 32s. Time estimates for 10 more iterations: 13m 20s, 100 more iterations: 2h 13m 21s, 500 more iterations: 11h 6m 45s. [2026-03-25 21:15:03,519][__main__][INFO] - Starting iteration 28. [2026-03-25 21:15:03,921][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:15:03,922][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:15:12,083][mllm.models.large_language_model_local][WARNING] - Response Proposal: 2 hats, 2 books, 16 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:15:16,438][__main__][INFO] - Number of regex retries in iteration 28: 1 [2026-03-25 21:15:16,439][__main__][INFO] - agents played in iteration 28 are Alice, Bob [2026-03-25 21:15:17,260][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:15:17,305][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:15:17,306][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:15:17,306][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:15:17,925][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:15:18,382][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:15:18,878][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:15:19,370][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:15:19,864][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:15:20,356][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:15:20,848][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:15:21,340][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:15:21,834][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:15:22,329][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:15:22,846][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:15:23,345][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:15:23,841][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:15:24,337][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:15:24,831][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:15:25,329][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:15:25,825][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:15:26,320][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:15:26,816][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:15:27,312][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:15:27,809][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:15:28,303][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:15:28,801][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:15:29,299][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:15:29,797][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:15:30,294][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:15:30,792][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:15:31,290][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:15:31,788][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:15:32,285][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:15:32,782][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:15:33,279][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:15:33,776][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:15:34,274][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:15:34,778][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:15:35,277][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:15:35,779][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:15:36,278][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:15:36,783][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:15:37,281][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:15:37,778][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:15:38,274][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:15:38,771][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:15:39,267][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:15:39,765][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:15:40,262][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:15:40,760][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:15:41,254][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:15:41,748][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:15:42,242][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:15:42,736][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:15:43,232][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:15:43,729][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:15:44,225][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:15:44,720][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:15:45,215][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:15:45,709][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:15:46,205][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:15:46,699][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:15:47,193][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:15:47,688][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:15:48,184][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:15:48,679][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:15:49,174][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:15:49,670][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:15:50,163][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:15:50,655][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:15:51,147][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:15:51,641][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:15:52,134][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:15:52,627][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:15:53,119][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:15:53,614][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:15:54,110][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:15:54,606][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:15:55,102][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:15:55,597][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:15:56,093][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:15:56,588][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:15:57,084][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:15:57,578][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:15:58,074][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:15:58,570][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:15:59,067][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:15:59,563][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:16:00,061][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:16:00,557][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:16:01,051][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:16:01,546][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:16:02,042][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:16:02,537][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:16:03,033][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:16:03,529][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:16:04,024][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:16:04,520][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:16:05,015][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:16:05,511][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:16:06,007][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:16:06,506][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:16:07,002][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:16:07,497][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:16:07,995][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:16:08,492][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:16:08,987][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:16:09,483][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:16:09,977][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:16:10,473][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:16:10,969][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:16:11,464][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:16:11,959][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:16:12,454][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:16:12,947][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:16:13,443][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:16:13,940][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:16:14,436][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:16:14,933][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:16:15,430][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:16:15,930][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:16:16,428][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:16:16,925][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:16:17,421][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:16:17,918][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:16:18,415][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:16:18,911][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:16:19,407][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:16:19,904][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:16:20,400][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:16:20,896][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:16:21,394][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20007 tokens. [2026-03-25 21:16:22,038][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.07%, Current % of VRAM taken: 60.43%, Block Peak % of device VRAM: 62.10%, ΔTime: 00:01:04 [2026-03-25 21:16:22,787][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:16:22,789][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:16:22,791][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:16:23,454][__main__][INFO] - Iteration 29 took 1m 19s (15.74% Gen, 83.43% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 34m 58s. Estimated total time: 66h 16m 39s. Time estimates for 10 more iterations: 13m 15s, 100 more iterations: 2h 12m 33s, 500 more iterations: 11h 2m 46s. [2026-03-25 21:16:23,456][__main__][INFO] - Starting iteration 29. [2026-03-25 21:16:23,869][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:16:23,870][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:16:25,418][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:16:26,695][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:16:37,067][__main__][INFO] - Number of regex retries in iteration 29: 2 [2026-03-25 21:16:37,068][__main__][INFO] - agents played in iteration 29 are Alice, Bob [2026-03-25 21:16:37,891][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.41%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:16:37,936][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.41%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:16:37,937][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:16:37,938][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:16:38,548][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:16:38,997][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:16:39,508][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:16:40,004][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:16:40,500][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:16:40,996][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:16:41,493][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:16:41,991][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:16:42,488][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:16:42,983][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:16:43,479][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:16:43,974][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:16:44,470][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:16:44,968][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:16:45,464][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:16:45,960][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:16:46,456][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:16:46,952][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:16:47,448][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:16:47,942][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:16:48,438][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:16:48,933][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:16:49,428][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:16:49,922][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:16:50,417][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:16:50,915][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:16:51,409][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:16:51,908][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:16:52,404][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:16:52,900][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:16:53,398][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:16:53,892][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:16:54,389][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:16:54,884][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:16:55,380][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:16:55,876][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:16:56,372][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:16:56,868][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:16:57,364][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:16:57,862][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:16:58,358][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:16:58,855][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:16:59,352][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:16:59,849][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:17:00,346][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:17:00,843][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:17:01,340][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:17:01,837][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:17:02,335][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:17:02,832][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:17:03,329][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:17:03,825][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:17:04,322][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:17:04,819][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:17:05,315][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:17:05,811][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:17:06,307][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:17:06,804][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:17:07,300][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:17:07,799][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:17:08,298][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:17:08,796][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:17:09,297][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:17:09,793][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:17:10,291][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:17:10,785][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:17:11,281][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:17:11,776][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:17:12,272][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:17:12,768][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:17:13,263][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:17:13,758][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:17:14,254][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:17:14,751][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:17:15,245][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:17:15,740][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:17:16,239][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:17:16,736][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:17:17,233][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:17:17,729][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:17:18,227][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:17:18,722][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:17:19,218][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:17:19,712][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:17:20,207][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:17:20,702][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:17:21,196][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:17:21,692][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:17:22,186][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:17:22,683][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:17:23,180][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:17:23,677][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:17:24,171][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:17:24,667][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:17:25,166][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:17:25,660][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:17:26,158][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:17:26,653][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:17:27,148][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:17:27,644][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:17:28,138][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:17:28,632][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:17:29,128][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:17:29,623][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:17:30,117][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:17:30,614][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:17:31,110][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:17:31,605][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:17:32,100][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:17:32,594][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:17:33,089][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:17:33,583][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:17:34,081][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:17:34,574][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:17:35,069][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:17:35,563][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:17:36,060][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:17:36,554][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:17:37,047][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:17:37,541][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:17:38,035][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:17:38,531][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:17:39,026][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:17:39,520][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:17:40,015][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:17:40,510][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:17:41,006][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:17:41,501][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:17:42,000][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20152 tokens. [2026-03-25 21:17:42,638][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.06%, Current % of VRAM taken: 60.41%, Block Peak % of device VRAM: 62.05%, ΔTime: 00:01:04 [2026-03-25 21:17:43,386][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:17:43,388][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:17:43,390][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:17:44,041][__main__][INFO] - Iteration 30 took 1m 20s (16.46% Gen, 82.72% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 5m 36s. Estimated total time: 66h 48m 38s. Time estimates for 10 more iterations: 13m 21s, 100 more iterations: 2h 13m 37s, 500 more iterations: 11h 8m 6s. [2026-03-25 21:17:44,043][__main__][INFO] - Starting iteration 30. [2026-03-25 21:17:44,455][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:17:44,456][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:17:51,458][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls This proposal takes advantage of the discrepancy in per-item values between you and Bob. You value hats the highest (10), while Bob values books and balls much higher (10 and 10, respectively). By taking all the hats, you ensure you receive the maximum possible value from this round, given your per-item value for hats is significantly higher than Bob's. This strategy aims to maximize your points in the short term by leveraging your value judgments. did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:17:59,322][__main__][INFO] - Number of regex retries in iteration 30: 1 [2026-03-25 21:17:59,323][__main__][INFO] - agents played in iteration 30 are Alice, Bob [2026-03-25 21:18:00,306][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:18:00,352][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:18:00,352][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:18:00,353][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:18:00,962][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:18:01,414][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:18:01,912][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:18:02,406][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:18:02,902][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:18:03,396][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:18:03,891][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:18:04,385][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:18:04,881][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:18:05,376][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:18:05,872][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:18:06,367][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:18:06,860][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:18:07,353][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:18:07,847][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:18:08,340][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:18:08,835][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:18:09,330][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:18:09,825][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:18:10,319][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:18:10,813][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:18:11,306][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:18:11,801][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:18:12,295][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:18:12,789][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:18:13,282][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:18:13,775][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:18:14,267][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:18:14,761][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:18:15,256][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:18:15,748][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:18:16,240][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:18:16,733][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:18:17,227][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:18:17,722][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:18:18,219][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:18:18,717][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:18:19,212][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:18:19,708][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:18:20,203][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:18:20,697][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:18:21,192][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:18:21,687][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:18:22,183][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:18:22,677][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:18:23,172][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:18:23,669][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:18:24,164][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:18:24,658][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:18:25,151][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:18:25,645][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:18:26,140][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:18:26,635][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:18:27,129][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:18:27,623][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:18:28,118][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:18:28,612][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:18:29,106][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:18:29,601][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:18:30,094][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:18:30,590][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:18:31,083][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:18:31,578][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:18:32,072][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:18:32,566][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:18:33,061][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:18:33,557][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:18:34,052][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:18:34,549][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:18:35,044][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:18:35,539][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:18:36,035][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:18:36,532][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:18:37,028][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:18:37,522][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:18:38,018][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:18:38,514][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:18:39,010][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:18:39,505][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:18:40,000][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:18:40,497][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:18:40,993][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:18:41,489][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:18:41,985][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:18:42,481][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:18:42,976][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:18:43,472][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:18:43,970][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:18:44,466][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:18:44,959][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:18:45,453][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:18:45,945][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:18:46,439][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:18:46,933][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:18:47,426][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:18:47,920][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:18:48,414][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:18:48,937][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:18:49,433][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:18:49,931][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:18:50,429][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:18:50,922][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:18:51,420][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:18:51,916][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:18:52,412][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:18:52,906][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:18:53,402][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:18:53,898][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:18:54,393][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:18:54,890][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:18:55,385][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:18:55,879][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:18:56,373][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:18:56,868][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:18:57,362][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:18:57,855][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:18:58,350][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:18:58,843][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:18:59,337][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:18:59,830][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:19:00,323][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:19:00,815][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:19:01,308][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:19:01,802][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:19:02,296][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:19:02,791][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:19:03,284][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:19:03,777][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:19:04,272][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19857 tokens. [2026-03-25 21:19:04,899][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.01%, Current % of VRAM taken: 60.37%, Block Peak % of device VRAM: 61.87%, ΔTime: 00:01:03 [2026-03-25 21:19:05,640][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:19:05,643][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:19:05,644][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:19:06,323][__main__][INFO] - Iteration 31 took 1m 21s (18.16% Gen, 81.01% Train). Generation: 14s, Training: 1m 6s. Estimated remaining time: 67h 29m 3s. Estimated total time: 68h 13m 27s. Time estimates for 10 more iterations: 13m 38s, 100 more iterations: 2h 16m 26s, 500 more iterations: 11h 22m 14s. [2026-03-25 21:19:06,326][__main__][INFO] - Starting iteration 31. [2026-03-25 21:19:06,724][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:19:06,725][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:19:07,257][mllm.models.large_language_model_local][WARNING] - Response Proposal: 5 hats, 0 books, 5 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:19:19,565][__main__][INFO] - Number of regex retries in iteration 31: 1 [2026-03-25 21:19:19,566][__main__][INFO] - agents played in iteration 31 are Alice, Bob [2026-03-25 21:19:20,572][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:19:20,617][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.39%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:19:20,618][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:19:20,618][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:19:21,237][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:19:21,690][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:19:22,196][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:19:22,695][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:19:23,193][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:19:23,691][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:19:24,190][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:19:24,688][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:19:25,186][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:19:25,681][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:19:26,177][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:19:26,673][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:19:27,171][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:19:27,668][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:19:28,164][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:19:28,660][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:19:29,156][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:19:29,654][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:19:30,153][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:19:30,648][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:19:31,141][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:19:31,636][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:19:32,131][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:19:32,627][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:19:33,122][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:19:33,616][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:19:34,110][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:19:34,603][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:19:35,099][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:19:35,592][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:19:36,086][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:19:36,580][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:19:37,074][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:19:37,569][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:19:38,064][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:19:38,560][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:19:39,054][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:19:39,553][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:19:40,048][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:19:40,541][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:19:41,036][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:19:41,532][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:19:42,026][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:19:42,521][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:19:43,016][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:19:43,513][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:19:44,010][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:19:44,504][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:19:44,998][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:19:45,492][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:19:45,984][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:19:46,476][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:19:46,969][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:19:47,462][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:19:47,956][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:19:48,452][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:19:48,944][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:19:49,439][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:19:49,935][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:19:50,429][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:19:50,923][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:19:51,418][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:19:51,914][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:19:52,409][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:19:52,905][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:19:53,400][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:19:53,897][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:19:54,394][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:19:54,892][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:19:55,388][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:19:55,884][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:19:56,381][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:19:56,876][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:19:57,371][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:19:57,863][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:19:58,357][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:19:58,851][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:19:59,345][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:19:59,838][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:20:00,334][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:20:00,827][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:20:01,322][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:20:01,816][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:20:02,311][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:20:02,809][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:20:03,305][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:20:03,800][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:20:04,296][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:20:04,792][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:20:05,287][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:20:05,783][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:20:06,279][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:20:06,776][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:20:07,271][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:20:07,769][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:20:08,265][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:20:08,761][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:20:09,257][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:20:09,755][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:20:10,252][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:20:10,748][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:20:11,244][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:20:11,740][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:20:12,236][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:20:12,734][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:20:13,233][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:20:13,731][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:20:14,230][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:20:14,726][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:20:15,221][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:20:15,717][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:20:16,212][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:20:16,707][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:20:17,199][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:20:17,693][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:20:18,191][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:20:18,684][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:20:19,177][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:20:19,671][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:20:20,164][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:20:20,657][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:20:21,152][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:20:21,648][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:20:22,142][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:20:22,637][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:20:23,132][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:20:23,627][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:20:24,122][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:20:24,617][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20017 tokens. [2026-03-25 21:20:25,256][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.08%, Current % of VRAM taken: 60.44%, Block Peak % of device VRAM: 61.91%, ΔTime: 00:01:04 [2026-03-25 21:20:26,026][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:20:26,029][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:20:26,030][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:20:26,765][__main__][INFO] - Iteration 32 took 1m 20s (16.04% Gen, 83.04% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 56m 19s. Estimated total time: 66h 42m 4s. Time estimates for 10 more iterations: 13m 20s, 100 more iterations: 2h 13m 24s, 500 more iterations: 11h 7m 0s. [2026-03-25 21:20:26,767][__main__][INFO] - Starting iteration 32. [2026-03-25 21:20:27,167][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:20:27,167][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:20:29,198][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:20:38,681][__main__][INFO] - Number of regex retries in iteration 32: 1 [2026-03-25 21:20:38,682][__main__][INFO] - agents played in iteration 32 are Alice, Bob [2026-03-25 21:20:39,488][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:20:39,533][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.40%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:20:39,534][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:20:39,535][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:20:40,153][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:20:40,605][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:20:41,104][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:20:41,600][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:20:42,095][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:20:42,591][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:20:43,086][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:20:43,581][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:20:44,076][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:20:44,570][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:20:45,063][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:20:45,560][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:20:46,056][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:20:46,551][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:20:47,048][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:20:47,543][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:20:48,038][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:20:48,538][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:20:49,035][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:20:49,533][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:20:50,032][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:20:50,529][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:20:51,030][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:20:51,529][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:20:52,050][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:20:52,546][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:20:53,043][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:20:53,539][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:20:54,038][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:20:54,535][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:20:55,032][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:20:55,531][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:20:56,028][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:20:56,525][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:20:57,021][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:20:57,519][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:20:58,017][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:20:58,515][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:20:59,008][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:20:59,500][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:20:59,995][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:21:00,491][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:21:00,991][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:21:01,486][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:21:01,982][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:21:02,479][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:21:02,974][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:21:03,470][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:21:03,970][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:21:04,462][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:21:04,957][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:21:05,452][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:21:05,947][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:21:06,441][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:21:06,936][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:21:07,431][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:21:07,926][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:21:08,425][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:21:08,919][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:21:09,415][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:21:09,909][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:21:10,403][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:21:10,898][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:21:11,396][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:21:11,894][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:21:12,389][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:21:12,883][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:21:13,379][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:21:13,874][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:21:14,367][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:21:14,861][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:21:15,356][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:21:15,850][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:21:16,344][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:21:16,837][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:21:17,331][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:21:17,825][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:21:18,320][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:21:18,815][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:21:19,311][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:21:19,805][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:21:20,300][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:21:20,795][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:21:21,292][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:21:21,788][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:21:22,284][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:21:22,781][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:21:23,276][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:21:23,772][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:21:24,266][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:21:24,761][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:21:25,255][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:21:25,752][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:21:26,247][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:21:26,741][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:21:27,237][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:21:27,733][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:21:28,227][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:21:28,721][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:21:29,215][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:21:29,707][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:21:30,200][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:21:30,695][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:21:31,190][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:21:31,685][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:21:32,181][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:21:32,683][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:21:33,180][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:21:33,678][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:21:34,178][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:21:34,674][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:21:35,172][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:21:35,673][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:21:36,169][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:21:36,664][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:21:37,160][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:21:37,657][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:21:38,154][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:21:38,650][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:21:39,145][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:21:39,642][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:21:40,139][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:21:40,634][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:21:41,132][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:21:41,629][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:21:42,126][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:21:42,622][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:21:43,120][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:21:43,615][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20101 tokens. [2026-03-25 21:21:44,261][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.11%, Current % of VRAM taken: 60.47%, Block Peak % of device VRAM: 62.07%, ΔTime: 00:01:04 [2026-03-25 21:21:45,048][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:21:45,050][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:21:45,052][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:21:45,735][__main__][INFO] - Iteration 33 took 1m 18s (14.66% Gen, 84.47% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 64h 41m 23s. Estimated total time: 65h 28m 27s. Time estimates for 10 more iterations: 13m 5s, 100 more iterations: 2h 10m 56s, 500 more iterations: 10h 54m 44s. [2026-03-25 21:21:45,738][__main__][INFO] - Starting iteration 33. [2026-03-25 21:21:46,138][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:21:46,139][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:21:55,215][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:21:58,306][__main__][INFO] - Number of regex retries in iteration 33: 1 [2026-03-25 21:21:58,307][__main__][INFO] - agents played in iteration 33 are Alice, Bob [2026-03-25 21:21:59,103][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:21:59,149][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:21:59,150][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:21:59,150][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:21:59,778][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:22:00,230][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:22:00,733][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:22:01,228][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:22:01,727][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:22:02,222][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:22:02,725][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:22:03,222][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:22:03,719][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:22:04,214][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:22:04,707][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:22:05,201][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:22:05,697][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:22:06,194][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:22:06,688][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:22:07,181][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:22:07,675][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:22:08,176][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:22:08,672][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:22:09,169][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:22:09,664][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:22:10,159][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:22:10,654][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:22:11,152][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:22:11,650][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:22:12,143][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:22:12,638][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:22:13,134][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:22:13,631][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:22:14,128][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:22:14,623][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:22:15,120][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:22:15,618][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:22:16,116][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:22:16,613][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:22:17,110][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:22:17,606][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:22:18,102][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:22:18,599][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:22:19,095][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:22:19,593][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:22:20,090][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:22:20,587][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:22:21,083][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:22:21,579][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:22:22,075][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:22:22,574][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:22:23,070][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:22:23,567][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:22:24,063][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:22:24,559][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:22:25,059][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:22:25,556][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:22:26,052][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:22:26,547][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:22:27,043][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:22:27,537][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:22:28,035][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:22:28,530][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:22:29,026][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:22:29,522][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:22:30,017][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:22:30,513][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:22:31,009][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:22:31,505][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:22:32,002][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:22:32,499][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:22:32,995][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:22:33,494][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:22:33,989][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:22:34,489][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:22:34,987][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:22:35,485][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:22:35,980][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:22:36,474][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:22:36,967][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:22:37,461][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:22:37,955][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:22:38,448][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:22:38,942][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:22:39,436][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:22:39,937][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:22:40,435][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:22:40,936][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:22:41,431][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:22:41,929][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:22:42,425][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:22:42,923][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:22:43,419][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:22:43,913][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:22:44,407][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:22:44,902][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:22:45,397][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:22:45,893][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:22:46,390][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:22:46,885][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:22:47,380][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:22:47,875][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:22:48,372][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:22:48,868][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:22:49,362][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:22:49,855][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:22:50,348][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:22:50,842][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:22:51,337][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:22:51,833][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:22:52,328][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:22:52,822][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:22:53,316][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:22:53,812][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:22:54,306][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:22:54,801][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:22:55,296][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:22:55,791][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:22:56,285][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:22:56,780][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:22:57,275][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:22:57,769][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:22:58,263][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:22:58,759][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:22:59,257][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:22:59,754][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:23:00,252][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:23:00,750][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:23:01,248][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:23:01,745][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:23:02,242][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:23:02,740][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:23:03,237][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20232 tokens. [2026-03-25 21:23:03,889][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.10%, Current % of VRAM taken: 60.45%, Block Peak % of device VRAM: 62.09%, ΔTime: 00:01:04 [2026-03-25 21:23:04,638][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:23:04,641][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:23:04,642][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:23:05,301][__main__][INFO] - Iteration 34 took 1m 19s (15.37% Gen, 83.80% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 9m 46s. Estimated total time: 65h 58m 9s. Time estimates for 10 more iterations: 13m 11s, 100 more iterations: 2h 11m 56s, 500 more iterations: 10h 59m 41s. [2026-03-25 21:23:05,303][__main__][INFO] - Starting iteration 34. [2026-03-25 21:23:05,722][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:23:05,722][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:23:18,337][__main__][INFO] - Number of regex retries in iteration 34: 0 [2026-03-25 21:23:18,338][__main__][INFO] - agents played in iteration 34 are Alice, Bob [2026-03-25 21:23:19,138][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:23:19,183][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:23:19,184][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:23:19,184][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:23:19,802][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:23:20,252][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:23:20,750][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:23:21,246][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:23:21,741][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:23:22,235][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:23:22,731][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:23:23,224][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:23:23,719][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:23:24,216][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:23:24,715][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:23:25,210][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:23:25,706][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:23:26,202][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:23:26,697][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:23:27,193][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:23:27,691][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:23:28,186][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:23:28,681][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:23:29,178][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:23:29,674][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:23:30,169][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:23:30,664][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:23:31,180][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:23:31,675][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:23:32,170][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:23:32,663][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:23:33,158][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:23:33,650][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:23:34,144][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:23:34,636][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:23:35,131][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:23:35,627][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:23:36,121][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:23:36,615][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:23:37,109][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:23:37,603][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:23:38,097][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:23:38,590][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:23:39,084][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:23:39,578][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:23:40,073][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:23:40,568][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:23:41,062][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:23:41,557][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:23:42,051][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:23:42,545][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:23:43,039][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:23:43,534][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:23:44,033][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:23:44,530][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:23:45,025][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:23:45,524][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:23:46,020][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:23:46,524][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:23:47,023][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:23:47,522][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:23:48,018][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:23:48,515][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:23:49,015][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:23:49,512][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:23:50,012][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:23:50,512][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:23:51,011][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:23:51,508][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:23:52,003][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:23:52,498][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:23:52,995][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:23:53,493][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:23:53,988][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:23:54,483][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:23:54,978][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:23:55,475][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:23:55,973][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:23:56,470][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:23:56,966][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:23:57,464][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:23:57,962][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:23:58,459][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:23:58,956][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:23:59,451][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:23:59,946][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:24:00,441][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:24:00,936][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:24:01,432][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:24:01,928][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:24:02,422][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:24:02,918][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:24:03,414][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:24:03,906][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:24:04,400][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:24:04,893][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:24:05,384][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:24:05,878][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:24:06,371][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:24:06,865][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:24:07,358][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:24:07,852][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:24:08,344][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:24:08,838][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:24:09,332][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:24:09,826][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:24:10,319][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:24:10,814][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:24:11,310][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:24:11,805][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:24:12,299][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:24:12,794][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:24:13,288][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:24:13,782][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:24:14,276][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:24:14,770][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:24:15,265][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:24:15,765][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:24:16,262][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:24:16,756][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:24:17,254][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:24:17,749][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:24:18,250][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:24:18,748][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:24:19,245][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:24:19,745][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:24:20,239][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:24:20,735][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:24:21,231][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:24:21,729][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:24:22,226][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:24:22,725][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:24:23,222][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 19978 tokens. [2026-03-25 21:24:23,864][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.10%, Current % of VRAM taken: 60.46%, Block Peak % of device VRAM: 62.17%, ΔTime: 00:01:04 [2026-03-25 21:24:24,604][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:24:24,609][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:24:24,610][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:24:25,261][__main__][INFO] - Iteration 35 took 1m 19s (15.86% Gen, 83.32% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 27m 15s. Estimated total time: 66h 16m 58s. Time estimates for 10 more iterations: 13m 15s, 100 more iterations: 2h 12m 33s, 500 more iterations: 11h 2m 49s. [2026-03-25 21:24:25,263][__main__][INFO] - Starting iteration 35. [2026-03-25 21:24:25,664][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:24:25,664][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:24:26,211][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:24:39,082][__main__][INFO] - Number of regex retries in iteration 35: 1 [2026-03-25 21:24:39,083][__main__][INFO] - agents played in iteration 35 are Alice, Bob [2026-03-25 21:24:40,087][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.42%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:24:40,133][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.42%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:24:40,134][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:24:40,134][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:24:40,752][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:24:41,204][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:24:41,708][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:24:42,206][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:24:42,707][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:24:43,203][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:24:43,702][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:24:44,200][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:24:44,699][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:24:45,196][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:24:45,694][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:24:46,192][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:24:46,688][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:24:47,184][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:24:47,680][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:24:48,177][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:24:48,674][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:24:49,172][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:24:49,670][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:24:50,169][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:24:50,665][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:24:51,162][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:24:51,667][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:24:52,165][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:24:52,661][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:24:53,156][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:24:53,653][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:24:54,152][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:24:54,646][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:24:55,142][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:24:55,638][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:24:56,134][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:24:56,629][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:24:57,124][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:24:57,619][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:24:58,116][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:24:58,612][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:24:59,107][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:24:59,603][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:25:00,099][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:25:00,594][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:25:01,092][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:25:01,587][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:25:02,082][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:25:02,578][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:25:03,075][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:25:03,574][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:25:04,069][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:25:04,564][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:25:05,059][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:25:05,556][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:25:06,053][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:25:06,549][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:25:07,045][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:25:07,541][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:25:08,037][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:25:08,532][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:25:09,027][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:25:09,522][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:25:10,018][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:25:10,515][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:25:11,013][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:25:11,508][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:25:12,005][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:25:12,500][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:25:12,996][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:25:13,493][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:25:13,988][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:25:14,487][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:25:14,980][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:25:15,480][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:25:15,979][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:25:16,476][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:25:16,972][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:25:17,467][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:25:17,963][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:25:18,459][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:25:18,959][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:25:19,456][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:25:19,952][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:25:20,451][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:25:20,948][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:25:21,447][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:25:21,946][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:25:22,445][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:25:22,942][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:25:23,442][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:25:23,957][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:25:24,457][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:25:24,955][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:25:25,452][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:25:25,952][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:25:26,448][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:25:26,946][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:25:27,443][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:25:27,943][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:25:28,439][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:25:28,936][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:25:29,434][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:25:29,933][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:25:30,431][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:25:30,927][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:25:31,422][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:25:31,918][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:25:32,413][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:25:32,910][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:25:33,405][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:25:33,902][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:25:34,400][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:25:34,895][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:25:35,394][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:25:35,889][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:25:36,386][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:25:36,880][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:25:37,377][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:25:37,873][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:25:38,369][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:25:38,863][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:25:39,361][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:25:39,857][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:25:40,353][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:25:40,847][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:25:41,343][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:25:41,838][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:25:42,336][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:25:42,830][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:25:43,323][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:25:43,818][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:25:44,316][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20375 tokens. [2026-03-25 21:25:44,943][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.19%, ΔTime: 00:01:04 [2026-03-25 21:25:45,732][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:25:45,735][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:25:45,736][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:25:46,485][__main__][INFO] - Iteration 36 took 1m 20s (16.60% Gen, 82.47% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 30m 0s. Estimated total time: 67h 21m 4s. Time estimates for 10 more iterations: 13m 28s, 100 more iterations: 2h 14m 42s, 500 more iterations: 11h 13m 30s. [2026-03-25 21:25:46,487][__main__][INFO] - Starting iteration 36. [2026-03-25 21:25:46,884][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:25:46,885][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:25:59,212][__main__][INFO] - Number of regex retries in iteration 36: 0 [2026-03-25 21:25:59,213][__main__][INFO] - agents played in iteration 36 are Alice, Bob [2026-03-25 21:26:00,018][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:26:00,063][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:26:00,064][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:26:00,064][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:26:00,681][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:26:01,133][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:26:01,629][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:26:02,125][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:26:02,622][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:26:03,119][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:26:03,615][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:26:04,111][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:26:04,607][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:26:05,105][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:26:05,604][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:26:06,102][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:26:06,600][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:26:07,102][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:26:07,600][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:26:08,098][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:26:08,600][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:26:09,097][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:26:09,595][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:26:10,094][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:26:10,591][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:26:11,088][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:26:11,585][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:26:12,082][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:26:12,578][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:26:13,076][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:26:13,575][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:26:14,074][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:26:14,574][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:26:15,070][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:26:15,564][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:26:16,061][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:26:16,556][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:26:17,050][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:26:17,544][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:26:18,039][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:26:18,536][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:26:19,032][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:26:19,526][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:26:20,022][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:26:20,517][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:26:21,013][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:26:21,508][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:26:22,001][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:26:22,495][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:26:22,991][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:26:23,485][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:26:23,981][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:26:24,476][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:26:24,971][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:26:25,469][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:26:25,964][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:26:26,461][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:26:26,958][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:26:27,452][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:26:27,949][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:26:28,466][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:26:28,959][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:26:29,453][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:26:29,946][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:26:30,440][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:26:30,935][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:26:31,431][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:26:31,925][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:26:32,419][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:26:32,913][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:26:33,406][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:26:33,899][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:26:34,394][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:26:34,888][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:26:35,381][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:26:35,874][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:26:36,368][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:26:36,863][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:26:37,359][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:26:37,854][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:26:38,353][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:26:38,851][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:26:39,346][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:26:39,842][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:26:40,341][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:26:40,838][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:26:41,333][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:26:41,829][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:26:42,323][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:26:42,818][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:26:43,313][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:26:43,808][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:26:44,303][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:26:44,798][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:26:45,294][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:26:45,794][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:26:46,291][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:26:46,785][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:26:47,284][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:26:47,781][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:26:48,278][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:26:48,775][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:26:49,272][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:26:49,768][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:26:50,264][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:26:50,763][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:26:51,262][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:26:51,762][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:26:52,258][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:26:52,753][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:26:53,247][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:26:53,743][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:26:54,237][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:26:54,734][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:26:55,230][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:26:55,726][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:26:56,222][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:26:56,720][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:26:57,220][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:26:57,718][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:26:58,220][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:26:58,719][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:26:59,217][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:26:59,715][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:27:00,212][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:27:00,707][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:27:01,200][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:27:01,695][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:27:02,189][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:27:02,683][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:27:03,177][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:27:03,671][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:27:04,164][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20248 tokens. [2026-03-25 21:27:04,811][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.08%, Current % of VRAM taken: 60.44%, Block Peak % of device VRAM: 62.09%, ΔTime: 00:01:04 [2026-03-25 21:27:05,551][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:27:05,553][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:27:05,554][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:27:06,254][__main__][INFO] - Iteration 37 took 1m 19s (15.53% Gen, 83.59% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 16m 6s. Estimated total time: 66h 8m 30s. Time estimates for 10 more iterations: 13m 13s, 100 more iterations: 2h 12m 17s, 500 more iterations: 11h 1m 25s. [2026-03-25 21:27:06,256][__main__][INFO] - Starting iteration 37. [2026-03-25 21:27:06,654][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:27:06,655][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:27:18,209][__main__][INFO] - Number of regex retries in iteration 37: 0 [2026-03-25 21:27:18,209][__main__][INFO] - agents played in iteration 37 are Alice, Bob [2026-03-25 21:27:19,011][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:27:19,056][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:27:19,056][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:27:19,057][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:27:19,669][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:27:20,127][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:27:20,629][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:27:21,126][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:27:21,624][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:27:22,122][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:27:22,619][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:27:23,116][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:27:23,615][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:27:24,110][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:27:24,605][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:27:25,101][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:27:25,597][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:27:26,094][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:27:26,589][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:27:27,084][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:27:27,580][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:27:28,077][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:27:28,575][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:27:29,071][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:27:29,570][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:27:30,064][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:27:30,559][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:27:31,057][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:27:31,555][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:27:32,053][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:27:32,548][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:27:33,044][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:27:33,541][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:27:34,037][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:27:34,534][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:27:35,028][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:27:35,525][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:27:36,022][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:27:36,518][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:27:37,015][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:27:37,512][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:27:38,008][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:27:38,504][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:27:39,002][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:27:39,499][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:27:39,999][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:27:40,500][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:27:40,997][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:27:41,497][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:27:41,996][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:27:42,497][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:27:42,997][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:27:43,495][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:27:43,991][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:27:44,484][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:27:44,979][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:27:45,474][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:27:45,970][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:27:46,464][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:27:46,959][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:27:47,454][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:27:47,954][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:27:48,453][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:27:48,951][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:27:49,446][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:27:49,944][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:27:50,439][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:27:50,937][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:27:51,432][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:27:51,934][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:27:52,434][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:27:52,933][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:27:53,430][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:27:53,928][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:27:54,426][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:27:54,923][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:27:55,425][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:27:55,920][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:27:56,416][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:27:56,912][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:27:57,407][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:27:57,903][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:27:58,398][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:27:58,894][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:27:59,389][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:27:59,884][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:28:00,382][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:28:00,878][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:28:01,376][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:28:01,871][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:28:02,367][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:28:02,862][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:28:03,357][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:28:03,851][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:28:04,344][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:28:04,837][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:28:05,331][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:28:05,825][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:28:06,318][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:28:06,812][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:28:07,304][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:28:07,799][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:28:08,294][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:28:08,789][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:28:09,284][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:28:09,779][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:28:10,274][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:28:10,771][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:28:11,266][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:28:11,761][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:28:12,260][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:28:12,754][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:28:13,250][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:28:13,745][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:28:14,241][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:28:14,739][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:28:15,239][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:28:15,736][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:28:16,235][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:28:16,733][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:28:17,231][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:28:17,728][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:28:18,226][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:28:18,723][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:28:19,220][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:28:19,721][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:28:20,220][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:28:20,721][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:28:21,221][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:28:21,718][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:28:22,217][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:28:22,714][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:28:23,212][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20180 tokens. [2026-03-25 21:28:23,849][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.14%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.17%, ΔTime: 00:01:04 [2026-03-25 21:28:24,593][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:28:24,596][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:28:24,597][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:28:25,249][__main__][INFO] - Iteration 38 took 1m 18s (14.70% Gen, 84.47% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 64h 36m 3s. Estimated total time: 65h 29m 46s. Time estimates for 10 more iterations: 13m 5s, 100 more iterations: 2h 10m 59s, 500 more iterations: 10h 54m 57s. [2026-03-25 21:28:25,251][__main__][INFO] - Starting iteration 38. [2026-03-25 21:28:25,660][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:28:25,661][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:28:26,315][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:28:38,486][__main__][INFO] - Number of regex retries in iteration 38: 1 [2026-03-25 21:28:38,487][__main__][INFO] - agents played in iteration 38 are Alice, Bob [2026-03-25 21:28:39,296][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:28:39,341][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:28:39,341][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:28:39,342][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:28:39,957][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:28:40,438][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:28:40,940][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:28:41,437][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:28:41,934][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:28:42,433][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:28:42,930][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:28:43,429][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:28:43,926][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:28:44,419][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:28:44,911][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:28:45,403][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:28:45,899][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:28:46,396][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:28:46,890][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:28:47,383][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:28:47,878][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:28:48,376][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:28:48,873][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:28:49,369][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:28:49,864][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:28:50,363][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:28:50,859][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:28:51,355][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:28:51,850][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:28:52,344][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:28:52,838][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:28:53,334][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:28:53,827][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:28:54,322][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:28:54,816][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:28:55,310][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:28:55,806][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:28:56,303][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:28:56,799][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:28:57,296][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:28:57,790][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:28:58,285][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:28:58,780][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:28:59,275][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:28:59,769][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:29:00,270][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:29:00,770][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:29:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:29:01,768][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:29:02,267][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:29:02,770][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:29:03,270][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:29:03,772][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:29:04,267][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:29:04,762][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:29:05,258][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:29:05,754][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:29:06,250][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:29:06,745][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:29:07,241][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:29:07,738][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:29:08,237][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:29:08,737][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:29:09,234][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:29:09,734][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:29:10,231][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:29:10,728][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:29:11,229][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:29:11,726][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:29:12,224][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:29:12,721][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:29:13,216][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:29:13,713][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:29:14,211][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:29:14,706][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:29:15,204][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:29:15,700][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:29:16,194][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:29:16,685][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:29:17,178][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:29:17,674][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:29:18,172][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:29:18,667][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:29:19,160][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:29:19,654][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:29:20,149][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:29:20,645][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:29:21,143][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:29:21,639][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:29:22,138][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:29:22,634][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:29:23,129][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:29:23,624][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:29:24,118][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:29:24,611][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:29:25,105][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:29:25,600][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:29:26,096][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:29:26,590][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:29:27,084][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:29:27,579][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:29:28,075][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:29:28,570][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:29:29,065][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:29:29,561][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:29:30,057][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:29:30,552][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:29:31,046][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:29:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:29:32,041][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:29:32,535][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:29:33,033][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:29:33,530][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:29:34,028][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:29:34,525][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:29:35,022][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:29:35,519][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:29:36,012][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:29:36,505][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:29:36,999][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:29:37,497][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:29:37,991][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:29:38,484][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:29:38,978][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:29:39,473][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:29:39,971][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:29:40,468][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:29:40,962][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:29:41,458][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:29:41,954][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:29:42,450][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:29:42,946][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:29:43,444][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20245 tokens. [2026-03-25 21:29:44,091][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.14%, Current % of VRAM taken: 60.49%, Block Peak % of device VRAM: 62.13%, ΔTime: 00:01:04 [2026-03-25 21:29:44,873][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:29:44,876][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:29:44,877][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:29:45,577][__main__][INFO] - Iteration 39 took 1m 19s (16.05% Gen, 83.07% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 40m 49s. Estimated total time: 66h 35m 53s. Time estimates for 10 more iterations: 13m 19s, 100 more iterations: 2h 13m 11s, 500 more iterations: 11h 5m 58s. [2026-03-25 21:29:45,579][__main__][INFO] - Starting iteration 39. [2026-03-25 21:29:45,992][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:29:45,993][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:29:59,282][__main__][INFO] - Number of regex retries in iteration 39: 0 [2026-03-25 21:29:59,283][__main__][INFO] - agents played in iteration 39 are Alice, Bob [2026-03-25 21:30:00,297][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:30:00,342][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:30:00,343][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:30:00,343][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:30:00,965][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:30:01,421][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:30:01,923][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:30:02,422][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:30:02,921][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:30:03,418][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:30:03,916][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:30:04,413][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:30:04,910][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:30:05,406][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:30:05,905][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:30:06,402][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:30:06,898][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:30:07,395][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:30:07,895][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:30:08,393][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:30:08,887][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:30:09,380][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:30:09,874][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:30:10,372][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:30:10,865][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:30:11,359][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:30:11,853][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:30:12,347][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:30:12,841][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:30:13,337][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:30:13,834][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:30:14,330][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:30:14,825][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:30:15,320][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:30:15,816][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:30:16,312][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:30:16,810][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:30:17,303][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:30:17,799][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:30:18,296][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:30:18,790][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:30:19,284][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:30:19,778][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:30:20,276][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:30:20,771][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:30:21,266][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:30:21,760][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:30:22,255][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:30:22,750][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:30:23,245][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:30:23,740][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:30:24,234][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:30:24,729][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:30:25,227][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:30:25,727][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:30:26,247][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:30:26,745][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:30:27,251][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:30:27,749][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:30:28,251][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:30:28,750][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:30:29,247][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:30:29,742][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:30:30,239][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:30:30,732][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:30:31,225][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:30:31,722][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:30:32,219][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:30:32,716][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:30:33,213][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:30:33,710][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:30:34,207][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:30:34,704][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:30:35,205][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:30:35,705][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:30:36,203][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:30:36,702][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:30:37,200][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:30:37,697][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:30:38,196][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:30:38,694][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:30:39,191][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:30:39,689][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:30:40,186][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:30:40,683][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:30:41,180][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:30:41,678][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:30:42,176][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:30:42,673][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:30:43,170][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:30:43,666][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:30:44,162][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:30:44,660][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:30:45,159][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:30:45,658][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:30:46,157][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:30:46,657][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:30:47,154][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:30:47,650][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:30:48,147][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:30:48,646][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:30:49,140][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:30:49,635][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:30:50,129][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:30:50,623][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:30:51,118][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:30:51,613][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:30:52,108][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:30:52,603][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:30:53,099][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:30:53,596][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:30:54,090][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:30:54,586][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:30:55,081][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:30:55,577][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:30:56,071][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:30:56,566][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:30:57,065][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:30:57,562][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:30:58,061][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:30:58,559][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:30:59,062][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:30:59,561][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:31:00,059][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:31:00,557][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:31:01,055][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:31:01,552][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:31:02,048][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:31:02,545][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:31:03,042][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:31:03,539][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:31:04,036][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:31:04,534][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20445 tokens. [2026-03-25 21:31:05,195][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 21:31:05,953][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:31:05,955][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:31:05,957][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:31:06,855][__main__][INFO] - Iteration 40 took 1m 20s (16.44% Gen, 82.45% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 26m 46s. Estimated total time: 67h 23m 11s. Time estimates for 10 more iterations: 13m 28s, 100 more iterations: 2h 14m 46s, 500 more iterations: 11h 13m 51s. [2026-03-25 21:31:06,857][__main__][INFO] - Starting iteration 40. [2026-03-25 21:31:07,257][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:31:07,257][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:31:07,800][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:31:20,078][__main__][INFO] - Number of regex retries in iteration 40: 1 [2026-03-25 21:31:20,078][__main__][INFO] - agents played in iteration 40 are Alice, Bob [2026-03-25 21:31:21,093][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:31:21,138][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:31:21,139][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:31:21,140][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:31:21,760][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:31:22,212][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:31:22,715][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:31:23,214][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:31:23,712][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:31:24,210][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:31:24,707][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:31:25,206][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:31:25,703][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:31:26,201][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:31:26,701][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:31:27,204][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:31:27,704][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:31:28,202][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:31:28,719][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:31:29,217][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:31:29,719][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:31:30,218][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:31:30,716][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:31:31,214][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:31:31,711][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:31:32,206][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:31:32,702][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:31:33,200][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:31:33,696][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:31:34,191][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:31:34,684][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:31:35,178][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:31:35,673][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:31:36,166][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:31:36,660][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:31:37,155][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:31:37,648][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:31:38,145][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:31:38,642][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:31:39,137][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:31:39,632][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:31:40,127][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:31:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:31:41,118][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:31:41,613][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:31:42,108][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:31:42,602][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:31:43,098][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:31:43,595][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:31:44,089][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:31:44,584][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:31:45,079][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:31:45,574][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:31:46,069][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:31:46,565][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:31:47,061][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:31:47,557][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:31:48,053][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:31:48,550][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:31:49,045][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:31:49,539][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:31:50,034][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:31:50,529][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:31:51,025][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:31:51,521][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:31:52,016][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:31:52,512][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:31:53,007][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:31:53,501][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:31:53,997][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:31:54,495][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:31:54,991][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:31:55,485][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:31:55,980][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:31:56,476][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:31:56,972][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:31:57,465][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:31:57,960][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:31:58,457][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:31:58,955][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:31:59,452][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:31:59,946][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:32:00,442][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:32:00,938][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:32:01,436][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:32:01,930][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:32:02,425][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:32:02,920][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:32:03,416][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:32:03,913][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:32:04,410][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:32:04,907][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:32:05,404][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:32:05,901][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:32:06,398][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:32:06,896][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:32:07,392][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:32:07,888][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:32:08,384][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:32:08,880][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:32:09,377][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:32:09,875][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:32:10,373][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:32:10,870][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:32:11,367][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:32:11,864][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:32:12,363][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:32:12,860][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:32:13,356][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:32:13,852][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:32:14,351][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:32:14,848][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:32:15,345][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:32:15,841][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:32:16,337][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:32:16,832][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:32:17,327][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:32:17,822][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:32:18,320][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:32:18,815][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:32:19,313][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:32:19,808][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:32:20,302][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:32:20,799][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:32:21,293][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:32:21,787][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:32:22,282][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:32:22,778][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:32:23,275][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:32:23,768][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:32:24,264][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:32:24,760][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:32:25,254][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20323 tokens. [2026-03-25 21:32:25,895][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.06%, Current % of VRAM taken: 60.41%, Block Peak % of device VRAM: 62.06%, ΔTime: 00:01:04 [2026-03-25 21:32:26,645][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:32:26,647][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:32:26,649][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:32:27,527][__main__][INFO] - Iteration 41 took 1m 20s (15.97% Gen, 82.93% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 55m 48s. Estimated total time: 66h 53m 33s. Time estimates for 10 more iterations: 13m 22s, 100 more iterations: 2h 13m 47s, 500 more iterations: 11h 8m 55s. [2026-03-25 21:32:27,529][__main__][INFO] - Starting iteration 41. [2026-03-25 21:32:27,934][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:32:27,935][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:32:40,472][__main__][INFO] - Number of regex retries in iteration 41: 0 [2026-03-25 21:32:40,473][__main__][INFO] - agents played in iteration 41 are Alice, Bob [2026-03-25 21:32:41,274][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:32:41,319][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.43%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:32:41,319][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:32:41,320][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:32:41,939][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:32:42,394][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:32:42,898][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:32:43,393][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:32:43,889][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:32:44,388][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:32:44,883][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:32:45,380][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:32:45,879][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:32:46,377][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:32:46,874][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:32:47,372][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:32:47,867][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:32:48,365][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:32:48,862][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:32:49,359][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:32:49,856][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:32:50,357][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:32:50,856][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:32:51,356][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:32:51,852][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:32:52,350][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:32:52,847][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:32:53,347][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:32:53,848][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:32:54,345][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:32:54,841][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:32:55,338][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:32:55,835][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:32:56,332][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:32:56,829][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:32:57,327][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:32:57,823][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:32:58,322][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:32:58,827][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:32:59,328][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:32:59,828][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:33:00,326][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:33:00,824][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:33:01,320][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:33:01,818][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:33:02,311][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:33:02,806][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:33:03,302][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:33:03,798][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:33:04,294][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:33:04,789][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:33:05,283][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:33:05,777][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:33:06,272][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:33:06,771][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:33:07,268][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:33:07,763][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:33:08,259][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:33:08,755][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:33:09,251][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:33:09,746][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:33:10,241][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:33:10,737][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:33:11,232][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:33:11,727][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:33:12,223][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:33:12,718][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:33:13,212][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:33:13,707][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:33:14,203][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:33:14,699][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:33:15,192][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:33:15,687][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:33:16,182][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:33:16,678][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:33:17,174][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:33:17,670][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:33:18,169][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:33:18,666][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:33:19,161][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:33:19,657][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:33:20,151][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:33:20,646][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:33:21,142][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:33:21,637][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:33:22,136][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:33:22,634][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:33:23,129][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:33:23,624][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:33:24,119][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:33:24,615][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:33:25,113][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:33:25,611][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:33:26,105][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:33:26,600][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:33:27,096][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:33:27,589][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:33:28,084][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:33:28,579][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:33:29,076][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:33:29,571][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:33:30,067][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:33:30,566][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:33:31,063][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:33:31,560][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:33:32,059][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:33:32,560][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:33:33,058][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:33:33,556][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:33:34,052][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:33:34,548][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:33:35,044][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:33:35,541][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:33:36,036][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:33:36,532][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:33:37,027][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:33:37,522][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:33:38,019][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:33:38,518][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:33:39,017][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:33:39,515][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:33:40,013][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:33:40,510][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:33:41,007][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:33:41,505][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:33:42,001][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:33:42,499][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:33:42,998][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:33:43,496][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:33:43,994][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:33:44,487][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:33:44,980][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:33:45,475][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20529 tokens. [2026-03-25 21:33:46,120][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.07%, Current % of VRAM taken: 60.43%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-25 21:33:46,869][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:33:46,871][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:33:46,872][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:33:47,717][__main__][INFO] - Iteration 42 took 1m 19s (15.72% Gen, 83.22% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 30m 4s. Estimated total time: 66h 29m 9s. Time estimates for 10 more iterations: 13m 17s, 100 more iterations: 2h 12m 58s, 500 more iterations: 11h 4m 51s. [2026-03-25 21:33:47,719][__main__][INFO] - Starting iteration 42. [2026-03-25 21:33:48,121][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:33:48,121][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:34:00,658][__main__][INFO] - Number of regex retries in iteration 42: 0 [2026-03-25 21:34:00,659][__main__][INFO] - agents played in iteration 42 are Alice, Bob [2026-03-25 21:34:01,457][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:34:01,502][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:34:01,503][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:34:01,503][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:34:02,119][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:34:02,576][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:34:03,077][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:34:03,575][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:34:04,070][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:34:04,566][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:34:05,061][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:34:05,558][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:34:06,057][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:34:06,560][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:34:07,059][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:34:07,558][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:34:08,074][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:34:08,571][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:34:09,070][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:34:09,570][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:34:10,068][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:34:10,563][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:34:11,060][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:34:11,557][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:34:12,056][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:34:12,550][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:34:13,048][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:34:13,546][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:34:14,042][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:34:14,537][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:34:15,032][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:34:15,529][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:34:16,027][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:34:16,525][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:34:17,023][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:34:17,520][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:34:18,018][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:34:18,520][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:34:19,019][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:34:19,518][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:34:20,018][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:34:20,517][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:34:21,015][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:34:21,512][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:34:22,010][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:34:22,507][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:34:23,003][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:34:23,499][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:34:23,994][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:34:24,487][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:34:24,982][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:34:25,478][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:34:25,974][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:34:26,470][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:34:26,968][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:34:27,466][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:34:27,960][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:34:28,453][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:34:28,948][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:34:29,442][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:34:29,936][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:34:30,431][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:34:30,927][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:34:31,423][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:34:31,921][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:34:32,418][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:34:32,914][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:34:33,408][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:34:33,907][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:34:34,401][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:34:34,896][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:34:35,391][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:34:35,887][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:34:36,382][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:34:36,879][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:34:37,374][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:34:37,873][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:34:38,373][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:34:38,871][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:34:39,368][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:34:39,869][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:34:40,367][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:34:40,865][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:34:41,363][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:34:41,861][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:34:42,357][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:34:42,853][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:34:43,350][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:34:43,845][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:34:44,341][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:34:44,838][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:34:45,334][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:34:45,830][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:34:46,325][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:34:46,820][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:34:47,314][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:34:47,810][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:34:48,308][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:34:48,804][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:34:49,300][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:34:49,797][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:34:50,293][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:34:50,789][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:34:51,286][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:34:51,781][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:34:52,277][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:34:52,770][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:34:53,266][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:34:53,762][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:34:54,257][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:34:54,751][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:34:55,245][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:34:55,740][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:34:56,233][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:34:56,727][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:34:57,221][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:34:57,715][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:34:58,208][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:34:58,703][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:34:59,198][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:34:59,689][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:35:00,185][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:35:00,682][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:35:01,177][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:35:01,673][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:35:02,170][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:35:02,666][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:35:03,164][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:35:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:35:04,161][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:35:04,658][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:35:05,154][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:35:05,651][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20127 tokens. [2026-03-25 21:35:06,294][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.16%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.25%, ΔTime: 00:01:04 [2026-03-25 21:35:07,044][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:35:07,047][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:35:07,048][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:35:07,874][__main__][INFO] - Iteration 43 took 1m 19s (15.72% Gen, 83.24% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 27m 16s. Estimated total time: 66h 27m 42s. Time estimates for 10 more iterations: 13m 17s, 100 more iterations: 2h 12m 55s, 500 more iterations: 11h 4m 37s. [2026-03-25 21:35:07,876][__main__][INFO] - Starting iteration 43. [2026-03-25 21:35:08,286][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:35:08,287][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:35:08,962][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:35:21,144][__main__][INFO] - Number of regex retries in iteration 43: 1 [2026-03-25 21:35:21,144][__main__][INFO] - agents played in iteration 43 are Alice, Bob [2026-03-25 21:35:21,942][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:35:21,987][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:35:21,988][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:35:21,988][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:35:22,599][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:35:23,049][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:35:23,551][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:35:24,050][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:35:24,547][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:35:25,043][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:35:25,541][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:35:26,037][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:35:26,532][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:35:27,030][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:35:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:35:28,030][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:35:28,530][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:35:29,025][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:35:29,523][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:35:30,042][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:35:30,540][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:35:31,038][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:35:31,533][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:35:32,029][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:35:32,525][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:35:33,021][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:35:33,521][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:35:34,017][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:35:34,512][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:35:35,008][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:35:35,507][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:35:36,003][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:35:36,502][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:35:36,998][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:35:37,494][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:35:37,994][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:35:38,491][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:35:38,989][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:35:39,487][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:35:39,985][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:35:40,484][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:35:40,981][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:35:41,482][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:35:41,980][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:35:42,477][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:35:42,973][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:35:43,472][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:35:43,966][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:35:44,462][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:35:44,959][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:35:45,455][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:35:45,953][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:35:46,448][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:35:46,944][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:35:47,443][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:35:47,942][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:35:48,440][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:35:48,939][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:35:49,438][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:35:49,937][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:35:50,437][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:35:50,935][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:35:51,434][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:35:51,931][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:35:52,430][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:35:52,931][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:35:53,429][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:35:53,926][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:35:54,426][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:35:54,923][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:35:55,421][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:35:55,919][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:35:56,417][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:35:56,916][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:35:57,416][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:35:57,915][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:35:58,413][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:35:58,913][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:35:59,411][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:35:59,909][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:36:00,408][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:36:00,905][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:36:01,403][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:36:01,901][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:36:02,400][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:36:02,894][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:36:03,390][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:36:03,884][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:36:04,380][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:36:04,875][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:36:05,374][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:36:05,869][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:36:06,364][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:36:06,859][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:36:07,356][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:36:07,852][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:36:08,347][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:36:08,842][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:36:09,338][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:36:09,833][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:36:10,328][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:36:10,827][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:36:11,325][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:36:11,823][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:36:12,321][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:36:12,818][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:36:13,317][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:36:13,811][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:36:14,308][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:36:14,803][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:36:15,300][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:36:15,798][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:36:16,293][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:36:16,789][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:36:17,287][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:36:17,782][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:36:18,282][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:36:18,778][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:36:19,273][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:36:19,770][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:36:20,264][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:36:20,761][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:36:21,257][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:36:21,753][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:36:22,248][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:36:22,744][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:36:23,239][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:36:23,734][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:36:24,231][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:36:24,729][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:36:25,226][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:36:25,721][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:36:26,219][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20508 tokens. [2026-03-25 21:36:26,863][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.16%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.19%, ΔTime: 00:01:04 [2026-03-25 21:36:27,601][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:36:27,603][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:36:27,605][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:36:28,242][__main__][INFO] - Iteration 44 took 1m 19s (16.08% Gen, 83.12% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 36m 2s. Estimated total time: 66h 37m 48s. Time estimates for 10 more iterations: 13m 19s, 100 more iterations: 2h 13m 15s, 500 more iterations: 11h 6m 18s. [2026-03-25 21:36:28,244][__main__][INFO] - Starting iteration 44. [2026-03-25 21:36:28,644][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:36:28,645][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:36:30,634][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:36:41,416][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:36:42,206][__main__][INFO] - Number of regex retries in iteration 44: 2 [2026-03-25 21:36:42,207][__main__][INFO] - agents played in iteration 44 are Alice, Bob [2026-03-25 21:36:43,216][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:36:43,262][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:36:43,262][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:36:43,263][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:36:43,896][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:36:44,355][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:36:44,858][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:36:45,359][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:36:45,859][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:36:46,359][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:36:46,860][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:36:47,361][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:36:47,862][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:36:48,364][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:36:48,867][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:36:49,369][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:36:49,869][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:36:50,367][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:36:50,866][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:36:51,367][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:36:51,865][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:36:52,366][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:36:52,865][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:36:53,364][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:36:53,864][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:36:54,363][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:36:54,862][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:36:55,362][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:36:55,862][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:36:56,361][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:36:56,860][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:36:57,358][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:36:57,859][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:36:58,359][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:36:58,860][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:36:59,359][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:36:59,858][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:37:00,353][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:37:00,849][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:37:01,345][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:37:01,841][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:37:02,340][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:37:02,837][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:37:03,333][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:37:03,829][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:37:04,325][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:37:04,822][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:37:05,320][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:37:05,818][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:37:06,315][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:37:06,814][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:37:07,310][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:37:07,807][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:37:08,303][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:37:08,802][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:37:09,299][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:37:09,796][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:37:10,295][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:37:10,791][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:37:11,287][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:37:11,783][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:37:12,278][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:37:12,773][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:37:13,269][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:37:13,764][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:37:14,260][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:37:14,758][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:37:15,255][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:37:15,754][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:37:16,253][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:37:16,752][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:37:17,250][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:37:17,749][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:37:18,249][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:37:18,748][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:37:19,248][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:37:19,746][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:37:20,244][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:37:20,750][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:37:21,251][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:37:21,750][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:37:22,251][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:37:22,752][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:37:23,253][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:37:23,753][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:37:24,252][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:37:24,753][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:37:25,255][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:37:25,755][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:37:26,257][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:37:26,756][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:37:27,257][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:37:27,758][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:37:28,259][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:37:28,760][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:37:29,261][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:37:29,761][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:37:30,262][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:37:30,764][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:37:31,266][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:37:31,768][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:37:32,268][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:37:32,763][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:37:33,258][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:37:33,752][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:37:34,248][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:37:34,746][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:37:35,241][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:37:35,736][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:37:36,232][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:37:36,729][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:37:37,225][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:37:37,721][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:37:38,218][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:37:38,714][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:37:39,210][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:37:39,705][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:37:40,202][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:37:40,699][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:37:41,195][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:37:41,690][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:37:42,185][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:37:42,682][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:37:43,179][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:37:43,674][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:37:44,171][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:37:44,666][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:37:45,161][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:37:45,657][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:37:46,155][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:37:46,649][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:37:47,145][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:37:47,643][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20604 tokens. [2026-03-25 21:37:48,290][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-25 21:37:49,039][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:37:49,041][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:37:49,043][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:37:49,863][__main__][INFO] - Iteration 45 took 1m 21s (16.70% Gen, 82.29% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 37m 48s. Estimated total time: 67h 40m 56s. Time estimates for 10 more iterations: 13m 32s, 100 more iterations: 2h 15m 21s, 500 more iterations: 11h 16m 49s. [2026-03-25 21:37:49,865][__main__][INFO] - Starting iteration 45. [2026-03-25 21:37:50,266][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:37:50,266][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:38:02,868][__main__][INFO] - Number of regex retries in iteration 45: 0 [2026-03-25 21:38:02,869][__main__][INFO] - agents played in iteration 45 are Alice, Bob [2026-03-25 21:38:03,676][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:38:03,721][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:38:03,721][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:38:03,722][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:38:04,338][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:38:04,792][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:38:05,299][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:38:05,798][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:38:06,297][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:38:06,798][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:38:07,299][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:38:07,799][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:38:08,300][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:38:08,799][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:38:09,298][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:38:09,797][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:38:10,299][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:38:10,797][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:38:11,300][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:38:11,799][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:38:12,298][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:38:12,796][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:38:13,293][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:38:13,793][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:38:14,290][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:38:14,788][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:38:15,286][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:38:15,786][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:38:16,283][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:38:16,779][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:38:17,278][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:38:17,773][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:38:18,273][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:38:18,770][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:38:19,267][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:38:19,762][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:38:20,259][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:38:20,755][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:38:21,251][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:38:21,746][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:38:22,241][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:38:22,737][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:38:23,232][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:38:23,726][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:38:24,223][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:38:24,725][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:38:25,223][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:38:25,723][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:38:26,221][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:38:26,719][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:38:27,218][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:38:27,717][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:38:28,215][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:38:28,711][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:38:29,206][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:38:29,702][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:38:30,200][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:38:30,695][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:38:31,190][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:38:31,685][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:38:32,181][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:38:32,679][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:38:33,174][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:38:33,669][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:38:34,167][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:38:34,664][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:38:35,160][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:38:35,655][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:38:36,153][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:38:36,647][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:38:37,144][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:38:37,640][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:38:38,136][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:38:38,635][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:38:39,130][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:38:39,626][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:38:40,122][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:38:40,618][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:38:41,116][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:38:41,612][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:38:42,111][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:38:42,607][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:38:43,105][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:38:43,601][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:38:44,097][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:38:44,592][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:38:45,086][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:38:45,584][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:38:46,082][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:38:46,578][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:38:47,074][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:38:47,569][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:38:48,065][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:38:48,564][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:38:49,064][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:38:49,562][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:38:50,062][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:38:50,559][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:38:51,059][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:38:51,556][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:38:52,058][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:38:52,554][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:38:53,050][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:38:53,547][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:38:54,045][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:38:54,541][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:38:55,038][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:38:55,534][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:38:56,031][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:38:56,533][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:38:57,033][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:38:57,533][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:38:58,033][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:38:58,532][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:38:59,033][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:38:59,535][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:39:00,037][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:39:00,536][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:39:01,030][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:39:01,526][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:39:02,022][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:39:02,517][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:39:03,012][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:39:03,508][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:39:04,003][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:39:04,500][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:39:04,996][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:39:05,492][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:39:05,991][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:39:06,488][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:39:06,986][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:39:07,484][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:39:07,981][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20536 tokens. [2026-03-25 21:39:08,622][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.13%, Current % of VRAM taken: 60.48%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 21:39:09,370][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:39:09,372][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:39:09,374][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:39:10,019][__main__][INFO] - Iteration 46 took 1m 19s (15.80% Gen, 83.39% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 23m 15s. Estimated total time: 66h 27m 43s. Time estimates for 10 more iterations: 13m 17s, 100 more iterations: 2h 12m 55s, 500 more iterations: 11h 4m 37s. [2026-03-25 21:39:10,022][__main__][INFO] - Starting iteration 46. [2026-03-25 21:39:10,423][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:39:10,423][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:39:22,964][__main__][INFO] - Number of regex retries in iteration 46: 0 [2026-03-25 21:39:22,964][__main__][INFO] - agents played in iteration 46 are Alice, Bob [2026-03-25 21:39:23,767][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.42%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:39:23,812][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.42%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:39:23,813][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:39:23,813][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:39:24,426][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:39:24,876][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:39:25,378][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:39:25,876][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:39:26,371][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:39:26,866][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:39:27,363][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:39:27,859][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:39:28,354][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:39:28,849][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:39:29,344][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:39:29,840][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:39:30,336][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:39:30,833][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:39:31,327][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:39:31,823][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:39:32,319][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:39:32,823][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:39:33,322][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:39:33,826][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:39:34,326][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:39:34,825][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:39:35,326][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:39:35,824][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:39:36,322][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:39:36,822][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:39:37,319][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:39:37,820][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:39:38,321][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:39:38,821][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:39:39,320][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:39:39,819][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:39:40,319][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:39:40,818][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:39:41,318][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:39:41,818][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:39:42,315][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:39:42,817][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:39:43,314][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:39:43,815][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:39:44,314][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:39:44,811][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:39:45,310][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:39:45,805][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:39:46,303][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:39:46,801][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:39:47,300][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:39:47,796][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:39:48,294][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:39:48,791][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:39:49,290][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:39:49,784][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:39:50,282][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:39:50,781][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:39:51,279][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:39:51,778][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:39:52,292][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:39:52,789][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:39:53,284][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:39:53,784][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:39:54,282][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:39:54,782][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:39:55,280][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:39:55,777][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:39:56,271][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:39:56,767][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:39:57,263][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:39:57,761][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:39:58,256][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:39:58,752][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:39:59,248][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:39:59,744][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:40:00,240][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:40:00,735][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:40:01,230][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:40:01,728][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:40:02,222][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:40:02,717][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:40:03,212][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:40:03,706][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:40:04,203][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:40:04,701][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:40:05,199][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:40:05,698][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:40:06,196][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:40:06,692][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:40:07,190][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:40:07,686][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:40:08,182][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:40:08,682][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:40:09,178][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:40:09,677][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:40:10,178][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:40:10,676][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:40:11,172][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:40:11,670][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:40:12,169][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:40:12,666][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:40:13,163][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:40:13,658][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:40:14,153][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:40:14,652][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:40:15,149][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:40:15,650][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:40:16,149][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:40:16,649][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:40:17,149][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:40:17,647][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:40:18,145][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:40:18,647][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:40:19,148][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:40:19,645][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:40:20,141][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:40:20,639][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:40:21,140][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:40:21,640][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:40:22,140][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:40:22,641][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:40:23,141][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:40:23,642][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:40:24,142][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:40:24,642][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:40:25,139][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:40:25,641][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:40:26,140][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:40:26,638][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:40:27,139][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:40:27,635][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:40:28,134][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20686 tokens. [2026-03-25 21:40:28,857][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.11%, Current % of VRAM taken: 60.46%, Block Peak % of device VRAM: 62.17%, ΔTime: 00:01:04 [2026-03-25 21:40:29,627][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:40:29,630][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:40:29,631][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:40:30,295][__main__][INFO] - Iteration 47 took 1m 19s (15.70% Gen, 83.47% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 27m 49s. Estimated total time: 66h 33m 37s. Time estimates for 10 more iterations: 13m 18s, 100 more iterations: 2h 13m 7s, 500 more iterations: 11h 5m 36s. [2026-03-25 21:40:30,298][__main__][INFO] - Starting iteration 47. [2026-03-25 21:40:30,701][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:40:30,702][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:40:31,286][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:40:35,213][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:40:37,607][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:40:43,299][__main__][INFO] - Number of regex retries in iteration 47: 3 [2026-03-25 21:40:43,300][__main__][INFO] - agents played in iteration 47 are Alice, Bob [2026-03-25 21:40:44,108][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:40:44,153][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:40:44,153][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:40:44,154][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:40:44,764][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:40:45,244][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:40:45,747][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:40:46,248][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:40:46,746][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:40:47,245][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:40:47,743][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:40:48,243][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:40:48,743][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:40:49,238][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:40:49,736][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:40:50,235][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:40:50,733][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:40:51,230][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:40:51,724][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:40:52,222][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:40:52,716][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:40:53,213][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:40:53,709][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:40:54,206][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:40:54,704][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:40:55,203][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:40:55,701][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:40:56,200][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:40:56,697][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:40:57,198][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:40:57,696][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:40:58,196][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:40:58,695][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:40:59,190][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:40:59,686][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:41:00,184][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:41:00,682][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:41:01,175][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:41:01,669][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:41:02,165][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:41:02,660][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:41:03,156][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:41:03,651][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:41:04,146][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:41:04,642][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:41:05,141][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:41:05,639][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:41:06,137][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:41:06,633][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:41:07,128][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:41:07,624][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:41:08,124][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:41:08,622][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:41:09,121][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:41:09,619][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:41:10,118][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:41:10,618][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:41:11,116][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:41:11,615][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:41:12,114][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:41:12,613][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:41:13,111][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:41:13,609][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:41:14,108][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:41:14,607][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:41:15,104][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:41:15,602][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:41:16,100][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:41:16,598][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:41:17,098][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:41:17,595][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:41:18,095][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:41:18,595][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:41:19,094][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:41:19,592][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:41:20,089][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:41:20,588][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:41:21,087][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:41:21,585][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:41:22,084][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:41:22,582][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:41:23,081][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:41:23,580][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:41:24,078][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:41:24,573][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:41:25,070][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:41:25,566][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:41:26,063][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:41:26,561][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:41:27,057][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:41:27,555][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:41:28,053][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:41:28,548][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:41:29,047][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:41:29,543][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:41:30,042][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:41:30,539][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:41:31,035][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:41:31,531][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:41:32,026][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:41:32,523][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:41:33,017][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:41:33,514][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:41:34,010][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:41:34,505][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:41:35,001][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:41:35,496][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:41:35,991][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:41:36,486][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:41:36,981][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:41:37,479][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:41:37,975][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:41:38,470][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:41:38,966][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:41:39,463][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:41:39,961][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:41:40,459][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:41:40,959][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:41:41,458][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:41:41,957][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:41:42,456][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:41:42,955][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:41:43,454][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:41:43,952][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:41:44,451][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:41:44,949][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:41:45,448][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:41:45,948][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:41:46,447][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:41:46,945][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:41:47,443][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:41:47,942][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:41:48,439][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20665 tokens. [2026-03-25 21:41:49,082][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 21:41:49,726][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:41:49,729][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:41:49,731][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:41:50,287][__main__][INFO] - Iteration 48 took 1m 19s (15.83% Gen, 83.47% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 12m 23s. Estimated total time: 66h 19m 31s. Time estimates for 10 more iterations: 13m 15s, 100 more iterations: 2h 12m 39s, 500 more iterations: 11h 3m 15s. [2026-03-25 21:41:50,290][__main__][INFO] - Starting iteration 48. [2026-03-25 21:41:50,689][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:41:50,690][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:42:02,550][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:42:03,305][__main__][INFO] - Number of regex retries in iteration 48: 1 [2026-03-25 21:42:03,306][__main__][INFO] - agents played in iteration 48 are Alice, Bob [2026-03-25 21:42:04,109][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:42:04,154][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:42:04,155][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:42:04,156][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:42:04,761][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:42:05,215][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:42:05,718][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:42:06,215][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:42:06,714][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:42:07,210][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:42:07,704][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:42:08,203][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:42:08,698][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:42:09,193][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:42:09,689][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:42:10,186][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:42:10,682][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:42:11,178][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:42:11,678][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:42:12,176][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:42:12,674][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:42:13,172][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:42:13,668][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:42:14,167][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:42:14,667][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:42:15,164][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:42:15,662][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:42:16,160][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:42:16,657][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:42:17,151][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:42:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:42:18,148][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:42:18,645][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:42:19,142][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:42:19,638][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:42:20,136][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:42:20,633][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:42:21,131][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:42:21,627][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:42:22,126][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:42:22,624][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:42:23,122][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:42:23,619][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:42:24,116][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:42:24,613][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:42:25,108][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:42:25,604][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:42:26,101][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:42:26,598][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:42:27,093][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:42:27,588][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:42:28,086][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:42:28,583][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:42:29,081][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:42:29,578][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:42:30,076][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:42:30,571][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:42:31,067][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:42:31,563][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:42:32,060][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:42:32,557][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:42:33,053][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:42:33,551][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:42:34,047][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:42:34,542][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:42:35,037][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:42:35,531][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:42:36,026][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:42:36,521][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:42:37,020][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:42:37,519][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:42:38,017][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:42:38,516][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:42:39,014][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:42:39,510][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:42:40,008][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:42:40,505][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:42:41,003][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:42:41,500][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:42:41,997][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:42:42,492][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:42:42,989][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:42:43,484][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:42:43,981][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:42:44,477][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:42:44,975][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:42:45,471][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:42:45,968][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:42:46,466][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:42:46,964][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:42:47,462][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:42:47,960][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:42:48,459][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:42:48,954][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:42:49,452][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:42:49,950][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:42:50,447][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:42:50,943][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:42:51,439][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:42:51,935][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:42:52,434][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:42:52,932][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:42:53,427][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:42:53,923][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:42:54,422][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:42:54,920][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:42:55,417][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:42:55,913][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:42:56,412][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:42:56,908][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:42:57,403][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:42:57,900][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:42:58,395][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:42:58,892][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:42:59,389][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:42:59,884][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:43:00,378][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:43:00,873][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:43:01,369][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:43:01,865][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:43:02,362][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:43:02,860][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:43:03,355][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:43:03,849][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:43:04,347][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:43:04,845][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:43:05,339][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:43:05,834][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:43:06,330][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:43:06,826][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:43:07,321][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:43:07,815][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:43:08,311][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20659 tokens. [2026-03-25 21:43:08,950][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.11%, Current % of VRAM taken: 60.46%, Block Peak % of device VRAM: 62.09%, ΔTime: 00:01:04 [2026-03-25 21:43:09,691][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:43:09,693][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:43:09,695][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:43:10,529][__main__][INFO] - Iteration 49 took 1m 19s (15.80% Gen, 83.15% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 23m 32s. Estimated total time: 66h 32m 0s. Time estimates for 10 more iterations: 13m 18s, 100 more iterations: 2h 13m 4s, 500 more iterations: 11h 5m 20s. [2026-03-25 21:43:10,532][__main__][INFO] - Starting iteration 49. [2026-03-25 21:43:10,936][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:43:10,936][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:43:11,586][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:43:23,683][__main__][INFO] - Number of regex retries in iteration 49: 1 [2026-03-25 21:43:23,684][__main__][INFO] - agents played in iteration 49 are Alice, Bob [2026-03-25 21:43:24,490][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:43:24,535][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:43:24,536][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:43:24,537][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:43:25,154][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:43:25,605][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:43:26,108][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:43:26,602][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:43:27,098][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:43:27,598][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:43:28,092][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:43:28,589][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:43:29,084][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:43:29,581][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:43:30,077][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:43:30,577][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:43:31,071][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:43:31,567][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:43:32,062][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:43:32,559][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:43:33,055][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:43:33,556][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:43:34,055][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:43:34,553][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:43:35,050][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:43:35,546][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:43:36,044][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:43:36,543][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:43:37,039][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:43:37,535][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:43:38,031][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:43:38,529][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:43:39,025][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:43:39,525][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:43:40,022][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:43:40,519][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:43:41,015][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:43:41,511][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:43:42,007][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:43:42,503][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:43:43,000][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:43:43,495][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:43:43,991][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:43:44,487][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:43:44,983][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:43:45,478][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:43:45,973][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:43:46,469][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:43:46,966][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:43:47,463][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:43:47,959][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:43:48,456][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:43:48,952][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:43:49,447][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:43:49,947][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:43:50,443][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:43:50,939][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:43:51,436][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:43:51,934][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:43:52,428][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:43:52,924][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:43:53,418][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:43:53,914][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:43:54,407][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:43:54,901][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:43:55,397][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:43:55,891][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:43:56,386][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:43:56,881][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:43:57,377][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:43:57,872][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:43:58,368][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:43:58,867][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:43:59,365][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:43:59,861][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:44:00,359][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:44:00,854][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:44:01,352][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:44:01,846][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:44:02,345][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:44:02,839][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:44:03,334][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:44:03,832][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:44:04,328][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:44:04,825][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:44:05,323][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:44:05,844][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:44:06,339][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:44:06,835][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:44:07,334][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:44:07,831][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:44:08,330][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:44:08,829][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:44:09,326][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:44:09,824][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:44:10,322][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:44:10,820][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:44:11,320][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:44:11,818][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:44:12,317][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:44:12,815][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:44:13,311][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:44:13,806][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:44:14,302][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:44:14,797][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:44:15,291][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:44:15,786][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:44:16,281][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:44:16,776][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:44:17,271][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:44:17,767][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:44:18,264][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:44:18,760][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:44:19,259][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:44:19,756][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:44:20,251][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:44:20,746][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:44:21,243][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:44:21,741][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:44:22,237][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:44:22,733][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:44:23,229][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:44:23,728][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:44:24,222][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:44:24,718][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:44:25,213][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:44:25,708][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:44:26,203][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:44:26,698][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:44:27,193][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:44:27,687][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:44:28,181][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:44:28,677][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20502 tokens. [2026-03-25 21:44:29,314][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.09%, Current % of VRAM taken: 60.44%, Block Peak % of device VRAM: 62.14%, ΔTime: 00:01:04 [2026-03-25 21:44:30,051][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:44:30,053][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:44:30,055][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:44:30,713][__main__][INFO] - Iteration 50 took 1m 19s (15.98% Gen, 83.19% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 19m 4s. Estimated total time: 66h 28m 52s. Time estimates for 10 more iterations: 13m 17s, 100 more iterations: 2h 12m 57s, 500 more iterations: 11h 4m 48s. [2026-03-25 21:44:30,715][__main__][INFO] - Starting iteration 50. [2026-03-25 21:44:31,115][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 0 and human policies 1. [2026-03-25 21:44:31,115][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:44:31,708][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:44:33,749][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:44:44,144][__main__][INFO] - Number of regex retries in iteration 50: 2 [2026-03-25 21:44:44,145][__main__][INFO] - agents played in iteration 50 are Alice, Bob [2026-03-25 21:44:44,947][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:44:44,992][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:44:44,993][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:44:44,993][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:44:45,624][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:44:46,076][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:44:46,577][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:44:47,072][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:44:47,566][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:44:48,064][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:44:48,563][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:44:49,058][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:44:49,553][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:44:50,053][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:44:50,571][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:44:51,068][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:44:51,565][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:44:52,063][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:44:52,561][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:44:53,059][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:44:53,555][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:44:54,055][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:44:54,554][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:44:55,052][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:44:55,549][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:44:56,048][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:44:56,547][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:44:57,046][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:44:57,543][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:44:58,037][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:44:58,532][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:44:59,026][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:44:59,520][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:45:00,015][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:45:00,512][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:45:01,008][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:45:01,506][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:45:02,003][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:45:02,499][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:45:02,993][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:45:03,491][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:45:03,989][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:45:04,484][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:45:04,977][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:45:05,473][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:45:05,975][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:45:06,474][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:45:06,973][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:45:07,472][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:45:08,004][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:45:08,504][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:45:09,007][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:45:09,507][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:45:10,004][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:45:10,501][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:45:10,995][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:45:11,491][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:45:11,986][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:45:12,482][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:45:12,978][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:45:13,472][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:45:13,968][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:45:14,465][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:45:14,963][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:45:15,458][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:45:15,956][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:45:16,451][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:45:16,947][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:45:17,444][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:45:17,939][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:45:18,435][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:45:18,929][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:45:19,424][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:45:19,920][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:45:20,416][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:45:20,910][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:45:21,406][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:45:21,905][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:45:22,402][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:45:22,898][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:45:23,396][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:45:23,891][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:45:24,386][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:45:24,886][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:45:25,382][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:45:25,881][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:45:26,379][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:45:26,877][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:45:27,373][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:45:27,871][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:45:28,368][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:45:28,865][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:45:29,363][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:45:29,858][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:45:30,354][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:45:30,850][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:45:31,346][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:45:31,842][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:45:32,338][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:45:32,833][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:45:33,328][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:45:33,823][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:45:34,319][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:45:34,813][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:45:35,309][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:45:35,807][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:45:36,304][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:45:36,802][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:45:37,297][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:45:37,795][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:45:38,293][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:45:38,792][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:45:39,289][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:45:39,789][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:45:40,288][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:45:40,790][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:45:41,289][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:45:41,785][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:45:42,282][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:45:42,776][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:45:43,271][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:45:43,767][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:45:44,262][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:45:44,756][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:45:45,251][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:45:45,747][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:45:46,246][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:45:46,744][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:45:47,240][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:45:47,740][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:45:48,238][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:45:48,736][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:45:49,233][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20769 tokens. [2026-03-25 21:45:49,867][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-25 21:45:50,602][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:45:50,605][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:45:50,606][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:45:51,840][__main__][INFO] - Iteration 51 took 1m 20s (16.14% Gen, 82.33% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 66h 5m 9s. Estimated total time: 67h 16m 18s. Time estimates for 10 more iterations: 13m 27s, 100 more iterations: 2h 14m 32s, 500 more iterations: 11h 12m 43s. [2026-03-25 21:45:51,843][__main__][INFO] - Starting iteration 51. [2026-03-25 21:45:52,240][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:45:52,241][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:46:04,760][__main__][INFO] - Number of regex retries in iteration 51: 0 [2026-03-25 21:46:04,761][__main__][INFO] - agents played in iteration 51 are Alice, Bob [2026-03-25 21:46:05,563][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:46:05,608][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:46:05,609][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:46:05,609][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:46:06,223][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:46:06,674][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:46:07,179][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:46:07,673][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:46:08,171][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:46:08,668][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:46:09,164][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:46:09,662][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:46:10,160][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:46:10,656][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:46:11,152][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:46:11,650][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:46:12,147][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:46:12,643][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:46:13,143][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:46:13,638][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:46:14,135][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:46:14,631][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:46:15,127][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:46:15,630][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:46:16,129][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:46:16,628][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:46:17,125][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:46:17,623][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:46:18,121][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:46:18,616][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:46:19,111][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:46:19,611][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:46:20,110][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:46:20,608][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:46:21,106][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:46:21,603][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:46:22,102][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:46:22,603][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:46:23,129][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:46:23,630][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:46:24,130][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:46:24,629][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:46:25,131][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:46:25,630][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:46:26,129][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:46:26,627][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:46:27,126][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:46:27,623][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:46:28,122][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:46:28,621][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:46:29,119][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:46:29,617][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:46:30,114][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:46:30,609][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:46:31,105][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:46:31,602][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:46:32,098][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:46:32,593][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:46:33,090][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:46:33,586][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:46:34,085][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:46:34,579][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:46:35,077][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:46:35,574][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:46:36,071][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:46:36,569][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:46:37,067][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:46:37,565][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:46:38,063][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:46:38,558][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:46:39,056][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:46:39,551][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:46:40,049][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:46:40,546][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:46:41,041][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:46:41,537][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:46:42,035][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:46:42,530][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:46:43,027][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:46:43,524][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:46:44,020][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:46:44,516][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:46:45,011][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:46:45,507][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:46:46,004][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:46:46,502][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:46:47,001][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:46:47,501][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:46:48,000][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:46:48,497][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:46:48,993][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:46:49,491][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:46:49,988][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:46:50,486][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:46:50,982][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:46:51,481][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:46:51,979][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:46:52,477][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:46:52,974][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:46:53,472][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:46:53,969][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:46:54,468][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:46:54,966][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:46:55,465][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:46:55,963][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:46:56,460][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:46:56,958][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:46:57,457][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:46:57,956][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:46:58,455][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:46:58,953][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:46:59,451][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:46:59,950][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:47:00,448][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:47:00,946][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:47:01,444][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:47:01,943][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:47:02,438][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:47:02,936][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:47:03,431][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:47:03,928][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:47:04,425][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:47:04,922][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:47:05,420][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:47:05,918][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:47:06,413][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:47:06,911][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:47:07,406][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:47:07,906][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:47:08,402][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:47:08,901][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:47:09,393][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:47:09,889][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20947 tokens. [2026-03-25 21:47:10,535][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.14%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 21:47:11,268][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:47:11,270][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:47:11,272][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:47:12,090][__main__][INFO] - Iteration 52 took 1m 19s (15.68% Gen, 83.30% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 19m 58s. Estimated total time: 66h 32m 28s. Time estimates for 10 more iterations: 13m 18s, 100 more iterations: 2h 13m 4s, 500 more iterations: 11h 5m 24s. [2026-03-25 21:47:12,092][__main__][INFO] - Starting iteration 52. [2026-03-25 21:47:12,492][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:47:12,493][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:47:24,709][__main__][INFO] - Number of regex retries in iteration 52: 0 [2026-03-25 21:47:24,710][__main__][INFO] - agents played in iteration 52 are Alice, Bob [2026-03-25 21:47:25,506][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:47:25,551][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:47:25,552][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:47:25,553][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:47:26,165][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:47:26,616][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:47:27,119][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:47:27,616][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:47:28,112][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:47:28,608][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:47:29,104][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:47:29,601][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:47:30,098][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:47:30,593][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:47:31,089][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:47:31,584][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:47:32,083][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:47:32,581][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:47:33,076][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:47:33,571][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:47:34,067][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:47:34,562][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:47:35,058][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:47:35,554][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:47:36,048][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:47:36,545][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:47:37,039][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:47:37,533][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:47:38,028][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:47:38,526][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:47:39,023][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:47:39,522][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:47:40,024][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:47:40,522][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:47:41,019][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:47:41,543][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:47:42,040][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:47:42,537][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:47:43,035][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:47:43,534][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:47:44,030][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:47:44,530][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:47:45,026][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:47:45,527][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:47:46,028][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:47:46,524][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:47:47,021][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:47:47,516][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:47:48,014][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:47:48,513][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:47:49,008][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:47:49,505][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:47:50,002][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:47:50,498][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:47:50,993][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:47:51,490][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:47:51,986][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:47:52,484][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:47:52,977][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:47:53,472][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:47:53,968][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:47:54,467][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:47:54,964][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:47:55,462][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:47:55,961][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:47:56,458][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:47:56,956][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:47:57,454][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:47:57,950][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:47:58,445][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:47:58,939][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:47:59,434][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:47:59,928][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:48:00,423][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:48:00,918][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:48:01,414][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:48:01,908][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:48:02,404][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:48:02,899][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:48:03,394][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:48:03,889][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:48:04,386][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:48:04,880][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:48:05,375][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:48:05,870][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:48:06,365][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:48:06,860][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:48:07,355][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:48:07,853][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:48:08,348][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:48:08,842][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:48:09,339][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:48:09,834][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:48:10,331][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:48:10,830][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:48:11,328][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:48:11,828][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:48:12,327][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:48:12,825][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:48:13,325][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:48:13,824][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:48:14,320][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:48:14,818][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:48:15,314][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:48:15,812][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:48:16,309][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:48:16,805][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:48:17,303][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:48:17,802][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:48:18,300][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:48:18,795][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:48:19,293][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:48:19,788][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:48:20,287][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:48:20,784][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:48:21,282][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:48:21,777][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:48:22,272][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:48:22,768][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:48:23,265][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:48:23,762][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:48:24,257][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:48:24,752][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:48:25,247][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:48:25,746][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:48:26,243][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:48:26,740][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:48:27,240][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:48:27,744][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:48:28,242][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:48:28,742][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:48:29,239][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:48:29,737][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20455 tokens. [2026-03-25 21:48:30,379][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-25 21:48:31,118][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:48:31,120][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:48:31,122][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:48:31,831][__main__][INFO] - Iteration 53 took 1m 19s (15.40% Gen, 83.71% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 64h 53m 10s. Estimated total time: 66h 6m 59s. Time estimates for 10 more iterations: 13m 13s, 100 more iterations: 2h 12m 13s, 500 more iterations: 11h 1m 9s. [2026-03-25 21:48:31,834][__main__][INFO] - Starting iteration 53. [2026-03-25 21:48:32,235][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:48:32,236][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:48:33,811][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:48:38,451][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:48:44,689][__main__][INFO] - Number of regex retries in iteration 53: 2 [2026-03-25 21:48:44,690][__main__][INFO] - agents played in iteration 53 are Alice, Bob [2026-03-25 21:48:45,496][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:48:45,541][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:48:45,541][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:48:45,542][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:48:46,156][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:48:46,611][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:48:47,114][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:48:47,613][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:48:48,112][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:48:48,612][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:48:49,110][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:48:49,608][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:48:50,107][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:48:50,606][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:48:51,105][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:48:51,603][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:48:52,102][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:48:52,600][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:48:53,098][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:48:53,592][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:48:54,089][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:48:54,591][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:48:55,091][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:48:55,591][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:48:56,090][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:48:56,589][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:48:57,089][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:48:57,589][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:48:58,088][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:48:58,587][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:48:59,086][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:48:59,584][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:49:00,084][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:49:00,583][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:49:01,084][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:49:01,585][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:49:02,085][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:49:02,584][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:49:03,084][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:49:03,582][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:49:04,081][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:49:04,580][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:49:05,079][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:49:05,577][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:49:06,075][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:49:06,572][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:49:07,068][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:49:07,568][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:49:08,063][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:49:08,559][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:49:09,056][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:49:09,554][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:49:10,053][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:49:10,548][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:49:11,045][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:49:11,544][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:49:12,040][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:49:12,539][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:49:13,036][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:49:13,532][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:49:14,028][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:49:14,527][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:49:15,025][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:49:15,525][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:49:16,024][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:49:16,524][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:49:17,022][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:49:17,521][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:49:18,017][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:49:18,517][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:49:19,015][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:49:19,513][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:49:20,011][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:49:20,511][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:49:21,010][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:49:21,510][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:49:22,008][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:49:22,507][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:49:23,006][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:49:23,505][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:49:24,003][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:49:24,503][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:49:25,002][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:49:25,500][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:49:25,995][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:49:26,495][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:49:26,994][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:49:27,493][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:49:27,992][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:49:28,491][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:49:28,992][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:49:29,491][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:49:29,990][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:49:30,489][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:49:30,989][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:49:31,488][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:49:31,988][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:49:32,488][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:49:32,988][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:49:33,488][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:49:33,987][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:49:34,486][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:49:34,985][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:49:35,483][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:49:35,984][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:49:36,483][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:49:36,982][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:49:37,480][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:49:37,978][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:49:38,477][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:49:38,972][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:49:39,471][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:49:39,968][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:49:40,467][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:49:40,965][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:49:41,464][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:49:41,962][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:49:42,457][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:49:42,956][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:49:43,451][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:49:43,948][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:49:44,444][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:49:44,941][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:49:45,439][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:49:45,934][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:49:46,431][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:49:46,930][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:49:47,428][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:49:47,925][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:49:48,425][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:49:48,924][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:49:49,420][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:49:49,916][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20939 tokens. [2026-03-25 21:49:50,559][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.13%, Current % of VRAM taken: 60.49%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-25 21:49:51,318][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:49:51,320][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:49:51,321][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:49:52,130][__main__][INFO] - Iteration 54 took 1m 19s (15.59% Gen, 83.40% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 19m 34s. Estimated total time: 66h 34m 44s. Time estimates for 10 more iterations: 13m 18s, 100 more iterations: 2h 13m 9s, 500 more iterations: 11h 5m 47s. [2026-03-25 21:49:52,132][__main__][INFO] - Starting iteration 54. [2026-03-25 21:49:52,546][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:49:52,547][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:49:53,229][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:50:05,529][__main__][INFO] - Number of regex retries in iteration 54: 1 [2026-03-25 21:50:05,530][__main__][INFO] - agents played in iteration 54 are Alice, Bob [2026-03-25 21:50:06,331][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:50:06,376][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:50:06,377][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:50:06,377][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:50:06,985][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:50:07,437][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:50:07,938][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:50:08,437][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:50:08,936][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:50:09,434][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:50:09,932][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:50:10,431][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:50:10,929][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:50:11,425][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:50:11,924][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:50:12,425][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:50:12,922][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:50:13,419][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:50:13,917][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:50:14,416][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:50:14,913][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:50:15,410][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:50:15,906][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:50:16,402][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:50:16,898][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:50:17,394][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:50:17,889][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:50:18,389][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:50:18,885][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:50:19,384][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:50:19,901][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:50:20,405][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:50:20,906][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:50:21,406][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:50:21,906][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:50:22,404][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:50:22,903][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:50:23,402][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:50:23,900][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:50:24,397][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:50:24,895][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:50:25,393][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:50:25,891][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:50:26,389][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:50:26,889][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:50:27,387][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:50:27,885][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:50:28,382][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:50:28,880][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:50:29,377][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:50:29,873][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:50:30,370][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:50:30,869][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:50:31,367][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:50:31,865][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:50:32,362][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:50:32,860][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:50:33,355][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:50:33,851][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:50:34,347][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:50:34,846][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:50:35,345][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:50:35,844][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:50:36,343][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:50:36,842][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:50:37,340][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:50:37,838][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:50:38,336][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:50:38,834][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:50:39,329][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:50:39,825][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:50:40,320][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:50:40,816][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:50:41,314][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:50:41,812][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:50:42,306][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:50:42,801][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:50:43,296][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:50:43,794][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:50:44,291][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:50:44,787][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:50:45,285][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:50:45,787][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:50:46,285][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:50:46,780][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:50:47,275][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:50:47,770][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:50:48,269][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:50:48,766][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:50:49,262][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:50:49,756][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:50:50,253][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:50:50,748][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:50:51,248][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:50:51,749][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:50:52,248][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:50:52,746][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:50:53,245][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:50:53,743][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:50:54,244][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:50:54,743][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:50:55,243][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:50:55,740][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:50:56,238][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:50:56,735][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:50:57,233][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:50:57,730][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:50:58,229][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:50:58,725][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:50:59,223][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:50:59,722][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:51:00,220][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:51:00,717][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:51:01,215][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:51:01,712][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:51:02,209][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:51:02,708][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:51:03,206][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:51:03,705][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:51:04,204][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:51:04,702][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:51:05,199][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:51:05,693][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:51:06,192][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:51:06,689][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:51:07,188][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:51:07,686][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:51:08,186][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:51:08,684][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:51:09,184][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:51:09,683][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:51:10,181][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:51:10,679][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21079 tokens. [2026-03-25 21:51:11,311][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 21:51:12,040][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:51:12,042][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:51:12,043][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:51:12,685][__main__][INFO] - Iteration 55 took 1m 20s (16.20% Gen, 83.00% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 30m 28s. Estimated total time: 66h 46m 59s. Time estimates for 10 more iterations: 13m 21s, 100 more iterations: 2h 13m 33s, 500 more iterations: 11h 7m 49s. [2026-03-25 21:51:12,687][__main__][INFO] - Starting iteration 55. [2026-03-25 21:51:13,086][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:51:13,087][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:51:13,659][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:51:22,407][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 ballsighetning did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:51:25,063][__main__][INFO] - Number of regex retries in iteration 55: 2 [2026-03-25 21:51:25,064][__main__][INFO] - agents played in iteration 55 are Alice, Bob [2026-03-25 21:51:25,869][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:51:25,914][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:51:25,914][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:51:25,915][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:51:26,540][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:51:26,996][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:51:27,500][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:51:28,000][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:51:28,500][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:51:29,000][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:51:29,498][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:51:29,996][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:51:30,495][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:51:30,990][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:51:31,487][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:51:31,985][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:51:32,483][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:51:32,981][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:51:33,477][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:51:33,975][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:51:34,469][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:51:34,968][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:51:35,464][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:51:35,963][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:51:36,458][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:51:36,954][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:51:37,450][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:51:37,946][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:51:38,445][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:51:38,943][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:51:39,441][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:51:39,937][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:51:40,436][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:51:40,933][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:51:41,432][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:51:41,929][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:51:42,430][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:51:42,928][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:51:43,423][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:51:43,919][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:51:44,417][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:51:44,912][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:51:45,410][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:51:45,907][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:51:46,404][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:51:46,904][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:51:47,400][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:51:47,897][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:51:48,397][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:51:48,895][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:51:49,394][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:51:49,892][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:51:50,391][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:51:50,890][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:51:51,389][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:51:51,887][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:51:52,385][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:51:52,884][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:51:53,382][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:51:53,882][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:51:54,382][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:51:54,878][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:51:55,375][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:51:55,872][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:51:56,369][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:51:56,866][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:51:57,363][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:51:57,859][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:51:58,356][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:51:58,854][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:51:59,351][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:51:59,851][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:52:00,351][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:52:00,850][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:52:01,348][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:52:01,847][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:52:02,346][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:52:02,844][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:52:03,342][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:52:03,840][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:52:04,335][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:52:04,834][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:52:05,331][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:52:05,829][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:52:06,326][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:52:06,824][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:52:07,323][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:52:07,822][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:52:08,317][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:52:08,817][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:52:09,315][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:52:09,812][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:52:10,310][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:52:10,809][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:52:11,307][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:52:11,806][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:52:12,305][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:52:12,804][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:52:13,304][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:52:13,802][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:52:14,301][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:52:14,797][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:52:15,292][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:52:15,792][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:52:16,289][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:52:16,786][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:52:17,282][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:52:17,779][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:52:18,276][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:52:18,775][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:52:19,273][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:52:19,770][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:52:20,267][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:52:20,763][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:52:21,262][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:52:21,756][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:52:22,254][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:52:22,756][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:52:23,256][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:52:23,754][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:52:24,253][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:52:24,752][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:52:25,250][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:52:25,750][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:52:26,250][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:52:26,747][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:52:27,242][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:52:27,739][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:52:28,235][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:52:28,732][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:52:29,229][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:52:29,726][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:52:30,224][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21022 tokens. [2026-03-25 21:52:30,861][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.33%, ΔTime: 00:01:04 [2026-03-25 21:52:31,593][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:52:31,596][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:52:31,597][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:52:32,326][__main__][INFO] - Iteration 56 took 1m 19s (15.12% Gen, 83.96% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 64h 44m 13s. Estimated total time: 66h 2m 3s. Time estimates for 10 more iterations: 13m 12s, 100 more iterations: 2h 12m 4s, 500 more iterations: 11h 0m 20s. [2026-03-25 21:52:32,329][__main__][INFO] - Starting iteration 56. [2026-03-25 21:52:32,730][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:52:32,730][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:52:33,317][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:52:33,393][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:52:45,185][__main__][INFO] - Number of regex retries in iteration 56: 2 [2026-03-25 21:52:45,185][__main__][INFO] - agents played in iteration 56 are Alice, Bob [2026-03-25 21:52:45,990][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:52:46,035][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:52:46,036][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:52:46,037][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:52:46,645][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:52:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:52:47,603][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:52:48,101][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:52:48,601][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:52:49,099][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:52:49,599][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:52:50,094][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:52:50,593][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:52:51,094][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:52:51,594][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:52:52,093][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:52:52,592][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:52:53,091][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:52:53,590][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:52:54,090][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:52:54,591][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:52:55,087][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:52:55,586][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:52:56,085][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:52:56,583][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:52:57,083][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:52:57,578][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:52:58,078][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:52:58,575][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:52:59,072][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:52:59,567][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:53:00,064][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:53:00,559][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:53:01,055][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:53:01,554][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:53:02,051][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:53:02,548][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:53:03,046][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:53:03,546][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:53:04,044][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:53:04,542][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:53:05,039][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:53:05,537][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:53:06,033][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:53:06,532][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:53:07,028][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:53:07,527][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:53:08,020][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:53:08,517][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:53:09,015][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:53:09,513][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:53:10,011][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:53:10,509][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:53:11,008][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:53:11,506][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:53:12,006][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:53:12,506][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:53:13,004][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:53:13,503][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:53:14,001][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:53:14,498][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:53:15,001][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:53:15,501][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:53:16,002][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:53:16,501][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:53:17,000][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:53:17,498][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:53:17,998][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:53:18,498][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:53:18,994][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:53:19,491][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:53:19,987][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:53:20,485][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:53:20,982][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:53:21,477][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:53:21,973][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:53:22,472][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:53:22,970][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:53:23,467][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:53:23,965][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:53:24,465][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:53:24,964][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:53:25,463][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:53:25,962][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:53:26,460][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:53:26,958][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:53:27,457][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:53:27,954][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:53:28,452][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:53:28,950][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:53:29,448][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:53:29,947][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:53:30,444][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:53:30,942][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:53:31,437][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:53:31,933][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:53:32,430][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:53:32,927][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:53:33,424][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:53:33,921][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:53:34,417][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:53:34,914][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:53:35,412][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:53:35,908][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:53:36,403][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:53:36,899][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:53:37,395][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:53:37,891][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:53:38,389][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:53:38,887][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:53:39,383][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:53:39,877][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:53:40,373][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:53:40,869][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:53:41,365][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:53:41,863][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:53:42,359][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:53:42,858][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:53:43,355][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:53:43,853][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:53:44,352][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:53:44,851][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:53:45,351][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:53:45,851][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:53:46,350][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:53:46,848][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:53:47,346][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:53:47,845][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:53:48,346][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:53:48,845][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:53:49,343][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:53:49,842][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:53:50,340][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20885 tokens. [2026-03-25 21:53:50,979][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 21:53:51,712][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:53:51,714][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:53:51,716][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:53:52,359][__main__][INFO] - Iteration 57 took 1m 19s (15.64% Gen, 83.55% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 2m 18s. Estimated total time: 66h 21m 28s. Time estimates for 10 more iterations: 13m 16s, 100 more iterations: 2h 12m 42s, 500 more iterations: 11h 3m 34s. [2026-03-25 21:53:52,361][__main__][INFO] - Starting iteration 57. [2026-03-25 21:53:52,762][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:53:52,763][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:54:05,516][__main__][INFO] - Number of regex retries in iteration 57: 0 [2026-03-25 21:54:05,517][__main__][INFO] - agents played in iteration 57 are Alice, Bob [2026-03-25 21:54:06,326][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:54:06,370][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:54:06,371][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:54:06,372][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:54:06,989][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:54:07,443][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:54:07,948][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:54:08,447][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:54:08,947][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:54:09,445][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:54:09,945][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:54:10,443][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:54:10,946][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:54:11,445][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:54:11,940][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:54:12,436][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:54:12,936][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:54:13,434][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:54:13,930][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:54:14,428][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:54:14,923][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:54:15,423][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:54:15,923][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:54:16,422][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:54:16,921][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:54:17,421][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:54:17,918][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:54:18,418][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:54:18,918][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:54:19,417][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:54:19,913][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:54:20,412][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:54:20,911][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:54:21,410][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:54:21,908][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:54:22,407][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:54:22,906][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:54:23,405][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:54:23,904][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:54:24,403][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:54:24,903][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:54:25,402][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:54:25,900][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:54:26,397][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:54:26,895][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:54:27,394][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:54:27,892][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:54:28,391][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:54:28,890][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:54:29,389][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:54:29,888][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:54:30,387][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:54:30,885][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:54:31,385][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:54:31,884][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:54:32,378][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:54:32,876][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:54:33,371][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:54:33,869][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:54:34,364][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:54:34,861][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:54:35,359][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:54:35,854][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:54:36,348][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:54:36,844][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:54:37,340][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:54:37,836][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:54:38,334][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:54:38,830][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:54:39,330][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:54:39,828][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:54:40,328][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:54:40,825][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:54:41,324][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:54:41,824][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:54:42,322][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:54:42,821][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:54:43,319][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:54:43,817][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:54:44,315][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:54:44,812][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:54:45,310][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:54:45,808][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:54:46,306][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:54:46,805][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:54:47,304][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:54:47,803][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:54:48,303][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:54:48,802][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:54:49,302][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:54:49,802][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:54:50,301][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:54:50,799][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:54:51,297][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:54:51,795][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:54:52,293][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:54:52,791][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:54:53,287][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:54:53,787][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:54:54,283][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:54:54,782][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:54:55,278][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:54:55,776][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:54:56,275][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:54:56,773][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:54:57,269][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:54:57,765][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:54:58,266][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:54:58,765][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:54:59,264][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:54:59,761][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:55:00,260][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:55:00,758][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:55:01,256][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:55:01,754][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:55:02,251][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:55:02,749][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:55:03,246][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:55:03,744][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:55:04,241][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:55:04,739][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:55:05,235][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:55:05,733][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:55:06,228][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:55:06,723][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:55:07,220][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:55:07,715][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:55:08,210][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:55:08,706][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:55:09,203][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:55:09,697][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:55:10,195][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:55:10,691][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21017 tokens. [2026-03-25 21:55:11,337][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.13%, Current % of VRAM taken: 60.48%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 21:55:12,136][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:55:13,025][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:55:13,026][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:55:13,850][__main__][INFO] - Iteration 58 took 1m 21s (15.73% Gen, 83.25% Train). Generation: 12s, Training: 1m 7s. Estimated remaining time: 66h 13m 53s. Estimated total time: 67h 34m 25s. Time estimates for 10 more iterations: 13m 30s, 100 more iterations: 2h 15m 8s, 500 more iterations: 11h 15m 44s. [2026-03-25 21:55:13,853][__main__][INFO] - Starting iteration 58. [2026-03-25 21:55:14,262][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:55:14,263][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:55:27,226][__main__][INFO] - Number of regex retries in iteration 58: 0 [2026-03-25 21:55:27,226][__main__][INFO] - agents played in iteration 58 are Alice, Bob [2026-03-25 21:55:28,031][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:55:28,076][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:55:28,077][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:55:28,077][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:55:28,690][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:55:29,141][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:55:29,643][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:55:30,141][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:55:30,636][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:55:31,136][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:55:31,634][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:55:32,130][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:55:32,626][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:55:33,126][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:55:33,622][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:55:34,121][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:55:34,619][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:55:35,119][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:55:35,615][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:55:36,112][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:55:36,610][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:55:37,113][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:55:37,613][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:55:38,135][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:55:38,634][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:55:39,133][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:55:39,632][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:55:40,131][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:55:40,630][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:55:41,128][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:55:41,626][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:55:42,123][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:55:42,622][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:55:43,119][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:55:43,614][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:55:44,113][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:55:44,611][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:55:45,106][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:55:45,606][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:55:46,105][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:55:46,604][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:55:47,102][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:55:47,599][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:55:48,098][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:55:48,597][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:55:49,095][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:55:49,593][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:55:50,089][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:55:50,586][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:55:51,084][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:55:51,582][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:55:52,079][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:55:52,576][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:55:53,073][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:55:53,571][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:55:54,068][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:55:54,563][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:55:55,061][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:55:55,559][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:55:56,057][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:55:56,554][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:55:57,052][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:55:57,549][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:55:58,048][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:55:58,543][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:55:59,041][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:55:59,539][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:56:00,038][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:56:00,534][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:56:01,029][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:56:01,523][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:56:02,018][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:56:02,513][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:56:03,009][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:56:03,506][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:56:04,001][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:56:04,495][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:56:04,995][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:56:05,489][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:56:05,987][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:56:06,484][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:56:06,984][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:56:07,482][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:56:07,979][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:56:08,477][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:56:08,976][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:56:09,473][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:56:09,971][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:56:10,469][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:56:10,968][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:56:11,465][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:56:11,963][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:56:12,463][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:56:12,959][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:56:13,455][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:56:13,954][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:56:14,452][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:56:14,949][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:56:15,447][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:56:15,944][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:56:16,442][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:56:16,937][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:56:17,435][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:56:17,934][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:56:18,434][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:56:18,934][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:56:19,432][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:56:19,929][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:56:20,427][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:56:20,925][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:56:21,424][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:56:21,922][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:56:22,421][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:56:22,920][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:56:23,418][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:56:23,917][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:56:24,416][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:56:24,913][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:56:25,412][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:56:25,911][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:56:26,410][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:56:26,908][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:56:27,406][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:56:27,903][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:56:28,401][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:56:28,896][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:56:29,394][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:56:29,890][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:56:30,390][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:56:30,888][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:56:31,386][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:56:31,884][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:56:32,379][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20942 tokens. [2026-03-25 21:56:33,013][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.22%, ΔTime: 00:01:04 [2026-03-25 21:56:33,746][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:56:33,748][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:56:33,750][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:56:34,424][__main__][INFO] - Iteration 59 took 1m 20s (16.17% Gen, 82.98% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 65h 26m 16s. Estimated total time: 66h 48m 8s. Time estimates for 10 more iterations: 13m 21s, 100 more iterations: 2h 13m 36s, 500 more iterations: 11h 8m 1s. [2026-03-25 21:56:34,427][__main__][INFO] - Starting iteration 59. [2026-03-25 21:56:34,825][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:56:34,826][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:56:47,270][__main__][INFO] - Number of regex retries in iteration 59: 0 [2026-03-25 21:56:47,271][__main__][INFO] - agents played in iteration 59 are Alice, Bob [2026-03-25 21:56:48,088][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:56:48,132][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:56:48,133][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:56:48,134][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:56:48,748][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:56:49,203][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:56:49,708][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:56:50,208][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:56:50,707][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:56:51,205][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:56:51,703][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:56:52,202][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:56:52,700][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:56:53,198][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:56:53,696][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:56:54,194][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:56:54,695][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:56:55,193][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:56:55,691][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:56:56,189][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:56:56,688][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:56:57,186][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:56:57,686][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:56:58,184][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:56:58,682][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:56:59,179][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:56:59,676][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:57:00,173][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:57:00,670][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:57:01,170][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:57:01,669][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:57:02,168][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:57:02,666][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:57:03,165][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:57:03,664][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:57:04,162][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:57:04,660][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:57:05,158][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:57:05,653][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:57:06,148][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:57:06,645][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:57:07,142][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:57:07,637][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:57:08,133][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:57:08,630][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:57:09,128][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:57:09,630][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:57:10,128][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:57:10,625][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:57:11,121][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:57:11,620][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:57:12,118][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:57:12,613][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:57:13,111][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:57:13,609][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:57:14,106][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:57:14,605][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:57:15,105][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:57:15,604][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:57:16,099][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:57:16,596][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:57:17,093][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:57:17,591][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:57:18,089][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:57:18,591][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:57:19,090][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:57:19,587][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:57:20,084][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:57:20,583][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:57:21,082][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:57:21,580][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:57:22,078][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:57:22,576][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:57:23,073][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:57:23,571][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:57:24,069][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:57:24,568][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:57:25,068][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:57:25,564][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:57:26,062][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:57:26,560][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:57:27,055][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:57:27,553][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:57:28,048][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:57:28,548][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:57:29,047][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:57:29,545][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:57:30,043][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:57:30,540][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:57:31,039][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:57:31,537][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:57:32,032][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:57:32,530][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:57:33,029][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:57:33,528][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:57:34,027][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:57:34,526][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:57:35,026][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:57:35,523][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:57:36,022][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:57:36,519][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:57:37,017][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:57:37,513][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:57:38,008][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:57:38,504][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:57:39,003][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:57:39,498][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:57:39,996][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:57:40,490][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:57:40,989][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:57:41,489][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:57:41,988][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:57:42,487][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:57:42,986][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:57:43,486][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:57:43,985][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:57:44,481][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:57:44,979][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:57:45,476][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:57:45,973][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:57:46,471][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:57:46,969][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:57:47,467][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:57:47,965][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:57:48,462][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:57:48,961][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:57:49,459][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:57:49,956][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:57:50,454][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:57:50,953][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:57:51,451][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:57:51,948][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:57:52,446][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21276 tokens. [2026-03-25 21:57:53,091][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.26%, ΔTime: 00:01:04 [2026-03-25 21:57:53,824][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:57:53,826][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:57:53,828][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:57:54,481][__main__][INFO] - Iteration 60 took 1m 19s (15.62% Gen, 83.55% Train). Generation: 12s, Training: 1m 6s. Estimated remaining time: 64h 59m 37s. Estimated total time: 66h 22m 49s. Time estimates for 10 more iterations: 13m 16s, 100 more iterations: 2h 12m 45s, 500 more iterations: 11h 3m 48s. [2026-03-25 21:57:54,483][__main__][INFO] - Starting iteration 60. [2026-03-25 21:57:54,885][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:57:54,886][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:57:57,076][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:58:06,681][__main__][INFO] - Number of regex retries in iteration 60: 1 [2026-03-25 21:58:06,682][__main__][INFO] - agents played in iteration 60 are Alice, Bob [2026-03-25 21:58:07,484][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:58:07,529][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:58:07,530][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:58:07,530][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:58:08,143][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:58:08,599][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:58:09,099][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:58:09,595][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:58:10,091][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:58:10,587][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:58:11,083][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:58:11,581][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:58:12,076][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:58:12,572][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:58:13,071][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:58:13,566][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:58:14,065][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:58:14,560][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:58:15,058][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:58:15,556][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:58:16,052][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:58:16,549][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:58:17,048][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:58:17,543][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:58:18,041][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:58:18,540][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:58:19,036][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:58:19,532][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:58:20,028][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:58:20,530][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:58:21,029][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:58:21,528][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:58:22,026][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:58:22,526][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:58:23,026][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:58:23,525][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:58:24,025][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:58:24,522][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:58:25,021][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:58:25,518][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:58:26,016][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:58:26,515][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:58:27,014][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:58:27,512][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:58:28,010][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:58:28,506][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:58:29,004][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:58:29,501][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:58:29,996][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:58:30,493][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:58:30,991][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:58:31,488][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:58:31,985][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:58:32,482][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:58:32,980][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:58:33,474][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:58:33,972][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:58:34,469][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:58:34,967][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:58:35,462][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:58:35,960][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:58:36,457][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:58:36,955][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:58:37,453][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:58:37,951][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:58:38,448][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:58:38,946][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 21:58:39,442][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 21:58:39,939][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 21:58:40,437][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 21:58:40,936][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 21:58:41,434][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 21:58:41,932][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 21:58:42,431][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 21:58:42,931][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 21:58:43,429][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 21:58:43,927][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 21:58:44,425][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 21:58:44,920][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 21:58:45,416][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 21:58:45,914][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 21:58:46,413][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 21:58:46,908][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 21:58:47,406][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 21:58:47,902][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 21:58:48,401][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 21:58:48,899][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 21:58:49,396][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 21:58:49,894][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 21:58:50,393][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 21:58:50,891][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 21:58:51,389][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 21:58:51,887][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 21:58:52,386][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 21:58:52,885][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 21:58:53,383][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 21:58:53,881][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 21:58:54,380][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 21:58:54,878][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 21:58:55,378][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 21:58:55,876][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 21:58:56,372][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 21:58:56,869][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 21:58:57,365][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 21:58:57,861][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 21:58:58,359][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 21:58:58,853][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 21:58:59,350][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 21:58:59,847][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 21:59:00,343][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 21:59:00,841][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 21:59:01,338][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 21:59:01,836][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 21:59:02,332][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 21:59:02,831][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 21:59:03,326][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 21:59:03,827][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 21:59:04,325][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 21:59:04,822][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 21:59:05,318][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 21:59:05,816][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 21:59:06,313][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 21:59:06,811][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 21:59:07,310][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 21:59:07,808][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 21:59:08,306][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 21:59:08,803][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 21:59:09,300][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 21:59:09,798][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 21:59:10,295][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 21:59:10,792][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 21:59:11,289][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 21:59:11,787][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21072 tokens. [2026-03-25 21:59:12,425][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.16%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 21:59:13,160][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 21:59:13,162][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 21:59:13,164][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 21:59:13,802][__main__][INFO] - Iteration 61 took 1m 18s (14.95% Gen, 84.24% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 64h 21m 23s. Estimated total time: 65h 45m 54s. Time estimates for 10 more iterations: 13m 9s, 100 more iterations: 2h 11m 31s, 500 more iterations: 10h 57m 39s. [2026-03-25 21:59:13,805][__main__][INFO] - Starting iteration 61. [2026-03-25 21:59:14,210][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 21:59:14,211][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 21:59:14,781][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:59:16,315][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:59:26,763][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls user Last Round Summary: - Items to split: 10 hats, 10 books, 10 balls - Your per-item values: hats=1, books=10, balls=10 - Bob's per-item values: hats=10, books=10, balls=1 - You proposed: 10 hats, 10 books, 10 balls - You earned: 51.0 points - Bob proposed: 10 hats, 10 books, 10 balls - Bob earned: 51.0 points - Round Complete. A New Round Begins The items to split are 10 hats, 10 books, 10 balls. Your per-item values are hats=10, books=10, balls=1 and Bob's per-item values are hats=1, books=1, balls=10. Submit Your Proposal Respond as Proposal: x hats, y books, z balls where x: 0-10 (integer), y: 0-10 (integer), z: 0-10 (integer). did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 21:59:27,522][__main__][INFO] - Number of regex retries in iteration 61: 3 [2026-03-25 21:59:27,523][__main__][INFO] - agents played in iteration 61 are Alice, Bob [2026-03-25 21:59:28,323][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:59:28,367][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 21:59:28,368][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 21:59:28,368][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 21:59:28,986][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 21:59:29,440][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 21:59:29,943][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 21:59:30,441][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 21:59:30,939][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 21:59:31,438][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 21:59:31,936][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 21:59:32,435][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 21:59:32,933][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 21:59:33,432][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 21:59:33,932][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 21:59:34,431][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 21:59:34,930][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 21:59:35,428][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 21:59:35,927][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 21:59:36,426][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 21:59:36,924][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 21:59:37,425][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 21:59:37,925][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 21:59:38,425][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 21:59:38,925][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 21:59:39,424][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 21:59:39,923][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 21:59:40,421][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 21:59:40,920][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 21:59:41,417][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 21:59:41,915][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 21:59:42,413][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 21:59:42,910][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 21:59:43,408][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 21:59:43,905][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 21:59:44,403][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 21:59:44,900][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 21:59:45,400][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 21:59:45,900][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 21:59:46,398][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 21:59:46,898][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 21:59:47,396][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 21:59:47,895][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 21:59:48,396][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 21:59:48,894][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 21:59:49,393][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 21:59:49,892][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 21:59:50,390][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 21:59:50,888][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 21:59:51,387][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 21:59:51,884][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 21:59:52,383][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 21:59:52,881][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 21:59:53,378][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 21:59:53,877][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 21:59:54,376][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 21:59:54,875][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 21:59:55,375][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 21:59:55,874][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 21:59:56,372][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 21:59:56,870][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 21:59:57,368][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 21:59:57,866][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 21:59:58,361][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 21:59:58,856][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 21:59:59,351][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 21:59:59,846][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:00:00,345][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:00:00,841][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:00:01,340][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:00:01,835][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:00:02,334][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:00:02,831][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:00:03,330][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:00:03,828][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:00:04,327][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:00:04,825][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:00:05,323][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:00:05,822][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:00:06,320][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:00:06,818][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:00:07,317][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:00:07,816][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:00:08,315][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:00:08,814][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:00:09,312][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:00:09,811][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:00:10,310][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:00:10,810][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:00:11,309][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:00:11,807][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:00:12,306][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:00:12,806][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:00:13,303][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:00:13,801][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:00:14,297][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:00:14,794][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:00:15,288][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:00:15,788][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:00:16,282][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:00:16,779][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:00:17,278][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:00:17,776][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:00:18,275][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:00:18,774][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:00:19,273][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:00:19,772][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:00:20,271][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:00:20,770][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:00:21,269][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:00:21,767][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:00:22,266][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:00:22,764][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:00:23,262][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:00:23,761][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:00:24,258][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:00:24,756][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:00:25,254][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:00:25,753][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:00:26,253][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:00:26,751][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:00:27,250][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:00:27,748][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:00:28,248][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:00:28,748][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:00:29,246][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:00:29,743][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:00:30,241][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:00:30,738][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:00:31,236][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:00:31,733][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:00:32,229][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:00:32,728][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21340 tokens. [2026-03-25 22:00:33,374][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:00:34,218][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:00:34,220][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:00:34,222][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:00:35,030][__main__][INFO] - Iteration 62 took 1m 20s (16.47% Gen, 82.53% Train). Generation: 13s, Training: 1m 6s. Estimated remaining time: 65h 55m 8s. Estimated total time: 67h 21m 1s. Time estimates for 10 more iterations: 13m 28s, 100 more iterations: 2h 14m 42s, 500 more iterations: 11h 13m 30s. [2026-03-25 22:00:35,033][__main__][INFO] - Starting iteration 62. [2026-03-25 22:00:35,434][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:00:35,434][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:00:36,522][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:00:46,820][__main__][INFO] - Number of regex retries in iteration 62: 1 [2026-03-25 22:00:46,821][__main__][INFO] - agents played in iteration 62 are Alice, Bob [2026-03-25 22:00:47,617][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:00:47,662][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:00:47,662][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:00:47,663][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:00:48,277][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:00:48,730][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:00:49,234][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:00:49,732][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:00:50,230][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:00:50,725][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:00:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:00:51,725][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:00:52,222][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:00:52,720][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:00:53,218][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:00:53,716][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:00:54,217][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:00:54,716][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:00:55,216][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:00:55,715][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:00:56,213][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:00:56,712][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:00:57,211][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:00:57,709][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:00:58,209][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:00:58,706][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:00:59,204][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:00:59,702][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:01:00,199][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:01:00,694][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:01:01,193][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:01:01,691][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:01:02,189][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:01:02,688][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:01:03,186][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:01:03,684][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:01:04,182][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:01:04,684][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:01:05,207][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:01:05,712][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:01:06,215][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:01:06,716][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:01:07,218][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:01:07,720][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:01:08,221][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:01:08,721][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:01:09,222][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:01:09,722][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:01:10,222][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:01:10,722][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:01:11,221][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:01:11,720][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:01:12,219][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:01:12,718][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:01:13,216][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:01:13,714][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:01:14,212][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:01:14,711][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:01:15,210][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:01:15,711][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:01:16,211][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:01:16,706][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:01:17,201][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:01:17,699][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:01:18,198][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:01:18,697][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:01:19,193][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:01:19,690][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:01:20,189][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:01:20,688][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:01:21,187][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:01:21,684][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:01:22,181][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:01:22,678][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:01:23,176][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:01:23,673][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:01:24,172][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:01:24,669][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:01:25,168][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:01:25,667][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:01:26,166][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:01:26,664][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:01:27,162][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:01:27,660][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:01:28,158][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:01:28,656][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:01:29,154][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:01:29,652][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:01:30,150][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:01:30,648][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:01:31,146][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:01:31,644][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:01:32,141][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:01:32,639][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:01:33,137][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:01:33,635][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:01:34,134][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:01:34,632][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:01:35,130][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:01:35,625][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:01:36,123][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:01:36,621][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:01:37,120][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:01:37,619][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:01:38,122][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:01:38,623][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:01:39,122][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:01:39,622][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:01:40,121][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:01:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:01:41,121][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:01:41,621][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:01:42,122][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:01:42,622][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:01:43,121][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:01:43,620][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:01:44,118][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:01:44,617][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:01:45,113][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:01:45,611][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:01:46,110][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:01:46,607][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:01:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:01:47,599][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:01:48,096][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:01:48,593][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:01:49,090][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:01:49,587][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:01:50,085][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:01:50,585][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:01:51,080][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:01:51,576][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:01:52,075][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21283 tokens. [2026-03-25 22:01:52,709][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-25 22:01:53,435][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:01:53,438][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:01:53,439][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:01:54,104][__main__][INFO] - Iteration 63 took 1m 18s (14.47% Gen, 84.68% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 64h 6m 19s. Estimated total time: 65h 33m 31s. Time estimates for 10 more iterations: 13m 6s, 100 more iterations: 2h 11m 7s, 500 more iterations: 10h 55m 35s. [2026-03-25 22:01:54,106][__main__][INFO] - Starting iteration 63. [2026-03-25 22:01:54,505][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:01:54,505][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:02:05,557][__main__][INFO] - Number of regex retries in iteration 63: 0 [2026-03-25 22:02:05,558][__main__][INFO] - agents played in iteration 63 are Alice, Bob [2026-03-25 22:02:06,355][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:02:06,399][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:02:06,400][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:02:06,401][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:02:07,014][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:02:07,469][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:02:07,972][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:02:08,471][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:02:08,970][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:02:09,469][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:02:09,969][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:02:10,468][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:02:10,964][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:02:11,462][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:02:11,962][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:02:12,461][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:02:12,959][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:02:13,458][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:02:13,957][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:02:14,456][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:02:14,955][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:02:15,454][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:02:15,951][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:02:16,450][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:02:16,948][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:02:17,443][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:02:17,942][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:02:18,438][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:02:18,937][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:02:19,437][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:02:19,936][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:02:20,436][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:02:20,936][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:02:21,435][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:02:21,935][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:02:22,434][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:02:22,933][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:02:23,431][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:02:23,928][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:02:24,425][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:02:24,922][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:02:25,420][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:02:25,917][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:02:26,415][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:02:26,911][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:02:27,410][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:02:27,908][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:02:28,405][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:02:28,904][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:02:29,402][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:02:29,898][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:02:30,396][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:02:30,894][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:02:31,393][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:02:31,891][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:02:32,385][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:02:32,882][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:02:33,378][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:02:33,875][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:02:34,373][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:02:34,870][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:02:35,371][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:02:35,870][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:02:36,369][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:02:36,871][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:02:37,373][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:02:37,872][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:02:38,372][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:02:38,872][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:02:39,370][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:02:39,868][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:02:40,368][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:02:40,865][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:02:41,364][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:02:41,861][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:02:42,359][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:02:42,857][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:02:43,356][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:02:43,856][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:02:44,354][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:02:44,853][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:02:45,353][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:02:45,854][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:02:46,354][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:02:46,853][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:02:47,350][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:02:47,846][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:02:48,343][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:02:48,841][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:02:49,335][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:02:49,833][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:02:50,331][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:02:50,825][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:02:51,327][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:02:51,826][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:02:52,325][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:02:52,824][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:02:53,322][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:02:53,820][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:02:54,318][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:02:54,816][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:02:55,315][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:02:55,810][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:02:56,306][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:02:56,805][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:02:57,302][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:02:57,795][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:02:58,293][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:02:58,787][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:02:59,285][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:02:59,782][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:03:00,279][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:03:00,776][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:03:01,274][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:03:01,771][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:03:02,269][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:03:02,767][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:03:03,263][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:03:03,760][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:03:04,255][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:03:04,753][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:03:05,250][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:03:05,747][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:03:06,246][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:03:06,744][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:03:07,243][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:03:07,742][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:03:08,240][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:03:08,739][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:03:09,239][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:03:09,737][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:03:10,237][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:03:10,736][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21288 tokens. [2026-03-25 22:03:11,378][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 22:03:12,109][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:03:12,112][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:03:12,113][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:03:12,760][__main__][INFO] - Iteration 64 took 1m 18s (14.12% Gen, 85.05% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 63h 44m 16s. Estimated total time: 65h 12m 47s. Time estimates for 10 more iterations: 13m 2s, 100 more iterations: 2h 10m 25s, 500 more iterations: 10h 52m 7s. [2026-03-25 22:03:12,762][__main__][INFO] - Starting iteration 64. [2026-03-25 22:03:13,163][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:03:13,164][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:03:13,766][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:03:14,689][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:03:24,222][__main__][INFO] - Number of regex retries in iteration 64: 2 [2026-03-25 22:03:24,223][__main__][INFO] - agents played in iteration 64 are Alice, Bob [2026-03-25 22:03:25,021][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:03:25,065][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:03:25,066][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:03:25,067][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:03:25,684][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:03:26,140][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:03:26,641][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:03:27,139][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:03:27,638][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:03:28,136][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:03:28,636][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:03:29,136][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:03:29,634][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:03:30,133][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:03:30,632][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:03:31,130][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:03:31,629][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:03:32,126][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:03:32,626][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:03:33,124][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:03:33,622][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:03:34,120][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:03:34,617][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:03:35,115][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:03:35,611][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:03:36,110][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:03:36,606][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:03:37,104][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:03:37,601][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:03:38,099][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:03:38,595][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:03:39,090][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:03:39,585][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:03:40,079][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:03:40,573][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:03:41,068][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:03:41,563][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:03:42,060][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:03:42,569][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:03:43,070][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:03:43,570][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:03:44,072][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:03:44,576][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:03:45,076][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:03:45,575][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:03:46,071][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:03:46,570][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:03:47,069][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:03:47,568][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:03:48,067][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:03:48,566][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:03:49,065][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:03:49,562][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:03:50,060][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:03:50,559][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:03:51,057][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:03:51,555][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:03:52,051][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:03:52,550][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:03:53,047][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:03:53,545][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:03:54,042][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:03:54,539][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:03:55,038][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:03:55,536][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:03:56,034][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:03:56,533][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:03:57,031][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:03:57,529][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:03:58,027][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:03:58,526][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:03:59,025][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:03:59,523][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:04:00,022][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:04:00,520][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:04:01,018][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:04:01,516][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:04:02,015][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:04:02,515][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:04:03,014][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:04:03,513][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:04:04,012][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:04:04,509][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:04:05,007][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:04:05,504][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:04:06,001][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:04:06,498][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:04:06,996][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:04:07,494][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:04:07,991][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:04:08,488][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:04:08,986][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:04:09,484][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:04:09,980][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:04:10,475][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:04:10,970][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:04:11,464][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:04:11,959][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:04:12,455][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:04:12,951][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:04:13,448][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:04:13,949][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:04:14,449][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:04:14,949][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:04:15,452][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:04:15,956][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:04:16,456][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:04:16,958][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:04:17,458][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:04:17,958][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:04:18,458][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:04:18,957][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:04:19,457][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:04:19,955][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:04:20,454][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:04:20,953][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:04:21,452][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:04:21,951][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:04:22,449][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:04:22,948][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:04:23,446][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:04:23,945][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:04:24,442][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:04:24,940][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:04:25,440][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:04:25,937][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:04:26,436][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:04:26,931][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:04:27,426][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:04:27,923][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:04:28,419][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:04:28,917][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:04:29,417][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21402 tokens. [2026-03-25 22:04:30,057][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-25 22:04:30,795][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:04:30,798][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:04:30,799][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:04:31,641][__main__][INFO] - Iteration 65 took 1m 18s (14.09% Gen, 84.83% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 63h 54m 5s. Estimated total time: 65h 23m 54s. Time estimates for 10 more iterations: 13m 4s, 100 more iterations: 2h 10m 47s, 500 more iterations: 10h 53m 59s. [2026-03-25 22:04:31,643][__main__][INFO] - Starting iteration 65. [2026-03-25 22:04:32,044][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:04:32,044][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:04:42,761][__main__][INFO] - Number of regex retries in iteration 65: 0 [2026-03-25 22:04:42,762][__main__][INFO] - agents played in iteration 65 are Alice, Bob [2026-03-25 22:04:43,567][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:04:43,612][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:04:43,612][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:04:43,613][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:04:44,223][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:04:44,678][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:04:45,178][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:04:45,675][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:04:46,173][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:04:46,672][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:04:47,170][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:04:47,668][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:04:48,168][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:04:48,669][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:04:49,168][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:04:49,666][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:04:50,165][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:04:50,663][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:04:51,161][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:04:51,661][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:04:52,159][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:04:52,663][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:04:53,178][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:04:53,678][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:04:54,178][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:04:54,679][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:04:55,178][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:04:55,677][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:04:56,178][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:04:56,676][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:04:57,174][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:04:57,669][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:04:58,170][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:04:58,668][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:04:59,168][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:04:59,667][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:05:00,165][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:05:00,663][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:05:01,160][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:05:01,658][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:05:02,155][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:05:02,652][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:05:03,150][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:05:03,647][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:05:04,146][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:05:04,643][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:05:05,141][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:05:05,639][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:05:06,137][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:05:06,635][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:05:07,134][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:05:07,632][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:05:08,130][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:05:08,628][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:05:09,125][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:05:09,622][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:05:10,119][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:05:10,616][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:05:11,114][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:05:11,612][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:05:12,110][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:05:12,609][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:05:13,109][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:05:13,608][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:05:14,107][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:05:14,606][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:05:15,105][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:05:15,604][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:05:16,102][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:05:16,600][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:05:17,098][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:05:17,596][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:05:18,095][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:05:18,594][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:05:19,093][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:05:19,590][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:05:20,089][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:05:20,587][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:05:21,086][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:05:21,583][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:05:22,081][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:05:22,576][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:05:23,075][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:05:23,570][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:05:24,066][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:05:24,569][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:05:25,497][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:05:25,995][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:05:26,495][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:05:26,995][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:05:27,494][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:05:27,992][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:05:28,491][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:05:28,991][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:05:29,490][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:05:29,987][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:05:30,486][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:05:30,984][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:05:31,482][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:05:31,980][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:05:32,478][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:05:32,977][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:05:33,475][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:05:33,973][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:05:34,472][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:05:34,970][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:05:35,467][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:05:35,966][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:05:36,464][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:05:36,963][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:05:37,461][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:05:37,959][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:05:38,457][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:05:38,952][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:05:39,452][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:05:39,950][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:05:40,444][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:05:40,943][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:05:41,440][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:05:41,938][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:05:42,436][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:05:42,934][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:05:43,431][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:05:43,929][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:05:44,426][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:05:44,926][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:05:45,428][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:05:45,931][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:05:46,430][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:05:46,928][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:05:47,427][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:05:47,925][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:05:48,426][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21334 tokens. [2026-03-25 22:05:49,060][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.39%, ΔTime: 00:01:04 [2026-03-25 22:05:49,807][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:05:49,809][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:05:49,811][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:05:50,470][__main__][INFO] - Iteration 66 took 1m 18s (13.67% Gen, 85.49% Train). Generation: 10s, Training: 1m 7s. Estimated remaining time: 63h 50m 10s. Estimated total time: 65h 21m 19s. Time estimates for 10 more iterations: 13m 4s, 100 more iterations: 2h 10m 42s, 500 more iterations: 10h 53m 33s. [2026-03-25 22:05:50,472][__main__][INFO] - Starting iteration 66. [2026-03-25 22:05:50,872][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:05:50,872][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:06:02,828][__main__][INFO] - Number of regex retries in iteration 66: 0 [2026-03-25 22:06:02,829][__main__][INFO] - agents played in iteration 66 are Alice, Bob [2026-03-25 22:06:03,639][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:06:03,683][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:06:03,684][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:06:03,685][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:06:04,289][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:06:04,744][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:06:05,247][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:06:05,746][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:06:06,245][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:06:06,743][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:06:07,241][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:06:07,739][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:06:08,237][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:06:08,736][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:06:09,235][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:06:09,733][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:06:10,232][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:06:10,733][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:06:11,231][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:06:11,729][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:06:12,228][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:06:12,728][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:06:13,230][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:06:13,731][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:06:14,230][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:06:14,729][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:06:15,244][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:06:15,744][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:06:16,244][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:06:16,742][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:06:17,237][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:06:17,736][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:06:18,237][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:06:18,735][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:06:19,235][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:06:19,732][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:06:20,231][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:06:20,727][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:06:21,226][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:06:21,721][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:06:22,216][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:06:22,712][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:06:23,209][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:06:23,705][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:06:24,201][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:06:24,699][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:06:25,199][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:06:25,697][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:06:26,196][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:06:26,694][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:06:27,192][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:06:27,690][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:06:28,190][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:06:28,689][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:06:29,188][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:06:29,686][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:06:30,181][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:06:30,679][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:06:31,177][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:06:31,675][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:06:32,172][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:06:32,671][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:06:33,169][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:06:33,667][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:06:34,166][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:06:34,664][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:06:35,161][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:06:35,658][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:06:36,156][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:06:36,655][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:06:37,153][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:06:37,651][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:06:38,149][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:06:38,647][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:06:39,145][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:06:39,643][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:06:40,140][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:06:40,638][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:06:41,137][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:06:41,637][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:06:42,135][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:06:42,634][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:06:43,133][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:06:43,632][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:06:44,131][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:06:44,630][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:06:45,130][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:06:45,629][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:06:46,128][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:06:46,629][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:06:47,127][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:06:47,626][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:06:48,124][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:06:48,623][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:06:49,119][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:06:49,615][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:06:50,111][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:06:50,610][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:06:51,108][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:06:51,602][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:06:52,096][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:06:52,595][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:06:53,093][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:06:53,591][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:06:54,091][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:06:54,590][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:06:55,088][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:06:55,586][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:06:56,084][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:06:56,582][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:06:57,080][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:06:57,576][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:06:58,071][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:06:58,566][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:06:59,065][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:06:59,563][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:07:00,061][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:07:00,557][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:07:01,056][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:07:01,552][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:07:02,051][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:07:02,550][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:07:03,049][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:07:03,546][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:07:04,044][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:07:04,542][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:07:05,040][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:07:05,536][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:07:06,035][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:07:06,534][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:07:07,033][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:07:07,531][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:07:08,030][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 20853 tokens. [2026-03-25 22:07:08,697][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-25 22:07:09,434][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:07:09,436][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:07:09,438][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:07:10,259][__main__][INFO] - Iteration 67 took 1m 19s (15.06% Gen, 83.90% Train). Generation: 11s, Training: 1m 6s. Estimated remaining time: 64h 36m 55s. Estimated total time: 66h 9m 23s. Time estimates for 10 more iterations: 13m 13s, 100 more iterations: 2h 12m 18s, 500 more iterations: 11h 1m 33s. [2026-03-25 22:07:10,261][__main__][INFO] - Starting iteration 67. [2026-03-25 22:07:10,660][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:07:10,660][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:07:21,257][__main__][INFO] - Number of regex retries in iteration 67: 0 [2026-03-25 22:07:21,258][__main__][INFO] - agents played in iteration 67 are Alice, Bob [2026-03-25 22:07:22,057][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:07:22,102][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:07:22,103][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:07:22,103][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:07:22,718][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:07:23,176][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:07:23,678][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:07:24,177][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:07:24,678][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:07:25,179][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:07:25,679][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:07:26,179][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:07:26,680][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:07:27,178][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:07:27,676][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:07:28,175][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:07:28,674][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:07:29,173][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:07:29,671][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:07:30,170][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:07:30,667][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:07:31,164][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:07:31,661][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:07:32,158][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:07:32,657][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:07:33,155][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:07:33,653][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:07:34,152][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:07:34,651][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:07:35,148][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:07:35,646][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:07:36,145][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:07:36,643][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:07:37,142][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:07:37,640][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:07:38,138][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:07:38,637][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:07:39,135][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:07:39,633][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:07:40,132][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:07:40,631][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:07:41,131][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:07:41,629][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:07:42,128][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:07:42,627][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:07:43,126][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:07:43,625][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:07:44,122][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:07:44,621][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:07:45,119][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:07:45,617][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:07:46,116][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:07:46,615][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:07:47,115][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:07:47,615][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:07:48,116][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:07:48,618][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:07:49,118][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:07:49,619][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:07:50,119][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:07:50,620][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:07:51,118][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:07:51,616][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:07:52,114][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:07:52,612][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:07:53,108][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:07:53,607][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:07:54,103][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:07:54,599][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:07:55,099][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:07:55,599][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:07:56,098][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:07:56,597][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:07:57,096][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:07:57,596][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:07:58,096][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:07:58,597][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:07:59,095][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:07:59,591][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:08:00,089][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:08:00,590][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:08:01,090][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:08:01,589][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:08:02,087][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:08:02,586][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:08:03,084][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:08:03,581][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:08:04,078][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:08:04,575][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:08:05,074][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:08:05,571][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:08:06,070][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:08:06,568][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:08:07,067][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:08:07,565][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:08:08,064][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:08:08,563][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:08:09,063][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:08:09,562][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:08:10,062][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:08:10,559][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:08:11,058][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:08:11,556][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:08:12,055][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:08:12,553][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:08:13,052][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:08:13,549][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:08:14,048][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:08:14,548][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:08:15,050][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:08:15,549][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:08:16,047][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:08:16,545][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:08:17,044][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:08:17,542][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:08:18,040][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:08:18,539][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:08:19,040][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:08:19,541][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:08:20,040][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:08:20,540][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:08:21,039][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:08:21,539][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:08:22,039][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:08:22,539][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:08:23,051][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:08:23,552][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:08:24,053][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:08:24,550][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:08:25,050][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:08:25,549][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:08:26,045][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:08:26,542][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21537 tokens. [2026-03-25 22:08:27,202][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-25 22:08:28,014][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:08:28,016][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:08:28,018][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:08:28,687][__main__][INFO] - Iteration 68 took 1m 18s (13.58% Gen, 85.56% Train). Generation: 10s, Training: 1m 6s. Estimated remaining time: 63h 27m 38s. Estimated total time: 65h 1m 24s. Time estimates for 10 more iterations: 13m 0s, 100 more iterations: 2h 10m 2s, 500 more iterations: 10h 50m 14s. [2026-03-25 22:08:28,690][__main__][INFO] - Starting iteration 68. [2026-03-25 22:08:29,091][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:08:29,092][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:08:29,666][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:08:30,379][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:08:33,504][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:08:39,299][__main__][INFO] - Number of regex retries in iteration 68: 3 [2026-03-25 22:08:39,300][__main__][INFO] - agents played in iteration 68 are Alice, Bob [2026-03-25 22:08:40,188][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:08:40,232][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:08:40,233][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:08:40,233][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:08:40,843][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:08:41,296][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:08:41,801][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:08:42,300][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:08:42,795][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:08:43,293][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:08:43,792][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:08:44,287][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:08:44,786][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:08:45,285][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:08:45,788][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:08:46,287][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:08:46,786][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:08:47,285][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:08:47,783][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:08:48,284][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:08:48,782][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:08:49,280][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:08:49,777][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:08:50,274][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:08:50,772][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:08:51,271][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:08:51,767][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:08:52,264][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:08:52,762][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:08:53,260][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:08:53,758][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:08:54,257][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:08:54,756][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:08:55,255][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:08:55,754][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:08:56,253][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:08:56,752][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:08:57,250][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:08:57,749][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:08:58,247][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:08:58,744][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:08:59,243][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:08:59,741][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:09:00,239][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:09:00,737][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:09:01,236][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:09:01,734][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:09:02,232][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:09:02,730][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:09:03,227][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:09:03,725][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:09:04,222][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:09:04,720][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:09:05,218][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:09:05,714][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:09:06,210][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:09:06,706][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:09:07,204][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:09:07,701][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:09:08,197][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:09:08,696][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:09:09,196][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:09:09,694][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:09:10,193][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:09:10,692][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:09:11,190][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:09:11,688][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:09:12,186][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:09:12,685][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:09:13,183][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:09:13,681][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:09:14,179][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:09:14,677][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:09:15,174][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:09:15,672][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:09:16,170][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:09:16,669][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:09:17,167][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:09:17,664][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:09:18,163][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:09:18,662][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:09:19,160][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:09:19,658][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:09:20,156][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:09:20,654][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:09:21,152][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:09:21,648][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:09:22,143][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:09:22,641][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:09:23,139][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:09:23,636][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:09:24,131][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:09:24,629][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:09:25,124][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:09:25,623][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:09:26,121][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:09:26,619][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:09:27,117][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:09:27,613][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:09:28,111][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:09:28,609][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:09:29,107][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:09:29,605][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:09:30,102][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:09:30,600][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:09:31,099][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:09:31,597][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:09:32,096][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:09:32,595][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:09:33,093][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:09:33,592][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:09:34,091][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:09:34,589][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:09:35,087][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:09:35,585][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:09:36,082][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:09:36,580][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:09:37,077][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:09:37,572][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:09:38,070][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:09:38,566][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:09:39,059][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:09:39,554][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:09:40,047][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:09:40,541][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:09:41,039][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:09:41,537][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:09:42,035][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:09:42,534][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:09:43,032][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:09:43,530][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:09:44,028][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:09:44,525][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21252 tokens. [2026-03-25 22:09:45,176][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-25 22:09:45,928][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:09:45,930][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:09:45,932][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:09:46,582][__main__][INFO] - Iteration 69 took 1m 17s (13.17% Gen, 85.99% Train). Generation: 10s, Training: 1m 6s. Estimated remaining time: 62h 59m 29s. Estimated total time: 64h 34m 34s. Time estimates for 10 more iterations: 12m 54s, 100 more iterations: 2h 9m 9s, 500 more iterations: 10h 45m 45s. [2026-03-25 22:09:46,585][__main__][INFO] - Starting iteration 69. [2026-03-25 22:09:46,985][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:09:46,985][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:09:57,230][__main__][INFO] - Number of regex retries in iteration 69: 0 [2026-03-25 22:09:57,231][__main__][INFO] - agents played in iteration 69 are Alice, Bob [2026-03-25 22:09:58,027][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:09:58,071][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:09:58,072][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:09:58,073][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:09:58,681][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:09:59,132][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:09:59,634][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:10:00,131][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:10:00,626][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:10:01,123][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:10:01,621][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:10:02,118][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:10:02,630][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:10:03,132][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:10:03,632][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:10:04,132][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:10:04,631][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:10:05,130][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:10:05,630][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:10:06,128][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:10:06,626][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:10:07,128][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:10:07,627][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:10:08,126][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:10:08,625][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:10:09,124][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:10:09,623][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:10:10,121][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:10:10,622][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:10:11,117][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:10:11,617][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:10:12,113][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:10:12,608][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:10:13,108][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:10:13,603][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:10:14,098][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:10:14,593][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:10:15,091][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:10:15,592][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:10:16,088][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:10:16,586][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:10:17,080][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:10:17,579][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:10:18,078][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:10:18,579][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:10:19,077][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:10:19,575][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:10:20,073][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:10:20,571][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:10:21,070][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:10:21,568][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:10:22,066][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:10:22,564][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:10:23,062][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:10:23,560][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:10:24,055][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:10:24,553][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:10:25,053][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:10:25,551][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:10:26,046][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:10:26,542][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:10:27,042][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:10:27,539][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:10:28,039][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:10:28,537][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:10:29,034][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:10:29,532][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:10:30,030][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:10:30,528][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:10:31,024][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:10:31,522][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:10:32,019][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:10:32,516][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:10:33,012][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:10:33,507][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:10:34,003][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:10:34,499][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:10:34,997][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:10:35,495][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:10:35,993][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:10:36,490][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:10:36,989][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:10:37,487][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:10:37,984][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:10:38,482][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:10:38,980][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:10:39,478][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:10:39,972][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:10:40,472][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:10:40,971][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:10:41,471][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:10:41,970][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:10:42,467][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:10:42,967][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:10:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:10:43,963][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:10:44,460][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:10:44,957][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:10:45,456][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:10:45,955][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:10:46,453][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:10:46,950][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:10:47,448][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:10:47,945][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:10:48,445][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:10:48,943][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:10:49,441][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:10:49,939][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:10:50,437][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:10:50,937][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:10:51,436][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:10:51,934][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:10:52,432][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:10:52,932][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:10:53,431][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:10:53,929][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:10:54,428][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:10:54,925][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:10:55,424][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:10:55,922][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:10:56,417][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:10:56,915][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:10:57,413][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:10:57,912][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:10:58,406][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:10:58,904][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:10:59,401][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:10:59,898][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:11:00,396][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:11:00,895][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:11:01,393][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:11:01,893][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:11:02,391][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21208 tokens. [2026-03-25 22:11:03,053][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-25 22:11:03,807][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:11:03,809][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:11:03,811][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:11:04,533][__main__][INFO] - Iteration 70 took 1m 17s (13.21% Gen, 85.85% Train). Generation: 10s, Training: 1m 6s. Estimated remaining time: 63h 1m 5s. Estimated total time: 64h 37m 27s. Time estimates for 10 more iterations: 12m 55s, 100 more iterations: 2h 9m 14s, 500 more iterations: 10h 46m 14s. [2026-03-25 22:11:04,535][__main__][INFO] - Starting iteration 70. [2026-03-25 22:11:04,937][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:11:04,938][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:11:14,699][__main__][INFO] - Number of regex retries in iteration 70: 0 [2026-03-25 22:11:14,700][__main__][INFO] - agents played in iteration 70 are Alice, Bob [2026-03-25 22:11:15,602][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:11:15,646][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:11:15,646][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:11:15,647][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:11:16,250][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:11:16,705][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:11:17,208][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:11:17,707][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:11:18,207][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:11:18,706][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:11:19,205][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:11:19,704][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:11:20,202][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:11:20,701][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:11:21,200][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:11:21,699][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:11:22,200][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:11:22,699][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:11:23,197][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:11:23,695][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:11:24,194][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:11:24,693][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:11:25,191][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:11:25,692][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:11:26,190][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:11:26,688][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:11:27,185][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:11:27,680][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:11:28,178][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:11:28,676][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:11:29,176][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:11:29,675][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:11:30,174][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:11:30,672][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:11:31,171][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:11:31,670][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:11:32,168][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:11:32,666][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:11:33,165][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:11:33,665][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:11:34,164][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:11:34,662][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:11:35,162][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:11:35,660][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:11:36,160][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:11:36,656][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:11:37,153][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:11:37,650][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:11:38,148][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:11:38,645][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:11:39,142][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:11:39,640][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:11:40,137][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:11:40,635][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:11:41,133][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:11:41,632][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:11:42,129][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:11:42,629][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:11:43,128][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:11:43,627][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:11:44,126][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:11:44,623][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:11:45,122][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:11:45,622][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:11:46,121][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:11:46,620][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:11:47,121][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:11:47,620][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:11:48,119][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:11:48,619][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:11:49,118][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:11:49,617][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:11:50,115][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:11:50,614][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:11:51,113][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:11:51,612][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:11:52,111][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:11:52,610][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:11:53,109][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:11:53,607][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:11:54,107][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:11:54,606][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:11:55,101][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:11:55,600][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:11:56,098][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:11:56,596][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:11:57,095][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:11:57,593][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:11:58,092][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:11:58,590][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:11:59,090][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:11:59,590][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:12:00,085][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:12:00,583][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:12:01,083][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:12:01,580][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:12:02,078][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:12:02,576][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:12:03,075][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:12:03,573][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:12:04,071][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:12:04,569][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:12:05,068][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:12:05,567][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:12:06,065][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:12:06,563][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:12:07,063][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:12:07,563][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:12:08,063][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:12:08,561][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:12:09,059][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:12:09,556][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:12:10,055][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:12:10,553][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:12:11,051][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:12:11,550][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:12:12,049][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:12:12,547][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:12:13,046][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:12:13,544][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:12:14,044][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:12:14,543][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:12:15,043][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:12:15,542][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:12:16,041][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:12:16,539][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:12:17,038][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:12:17,537][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:12:18,037][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:12:18,537][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:12:19,036][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:12:19,535][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:12:20,034][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21419 tokens. [2026-03-25 22:12:20,690][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:12:21,459][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:12:21,461][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:12:21,463][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:12:22,315][__main__][INFO] - Iteration 71 took 1m 17s (12.62% Gen, 86.28% Train). Generation: 9s, Training: 1m 6s. Estimated remaining time: 62h 51m 16s. Estimated total time: 64h 28m 57s. Time estimates for 10 more iterations: 12m 53s, 100 more iterations: 2h 8m 57s, 500 more iterations: 10h 44m 49s. [2026-03-25 22:12:22,318][__main__][INFO] - Starting iteration 71. [2026-03-25 22:12:22,723][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:12:22,723][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:12:32,330][__main__][INFO] - Number of regex retries in iteration 71: 0 [2026-03-25 22:12:32,331][__main__][INFO] - agents played in iteration 71 are Alice, Bob [2026-03-25 22:12:33,212][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:12:33,257][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:12:33,257][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:12:33,258][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:12:33,875][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:12:34,330][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:12:34,829][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:12:35,328][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:12:35,827][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:12:36,322][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:12:36,819][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:12:37,316][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:12:37,816][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:12:38,314][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:12:38,816][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:12:39,315][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:12:39,814][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:12:40,313][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:12:40,813][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:12:41,312][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:12:41,811][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:12:42,308][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:12:42,804][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:12:43,301][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:12:43,799][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:12:44,295][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:12:44,794][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:12:45,292][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:12:45,790][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:12:46,288][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:12:46,786][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:12:47,283][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:12:47,781][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:12:48,279][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:12:48,778][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:12:49,277][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:12:49,776][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:12:50,278][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:12:50,777][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:12:51,276][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:12:51,776][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:12:52,277][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:12:52,776][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:12:53,275][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:12:53,773][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:12:54,272][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:12:54,770][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:12:55,269][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:12:55,766][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:12:56,264][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:12:56,762][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:12:57,260][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:12:57,758][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:12:58,256][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:12:58,755][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:12:59,255][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:12:59,753][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:13:00,250][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:13:00,749][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:13:01,248][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:13:01,746][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:13:02,244][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:13:02,743][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:13:03,241][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:13:03,740][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:13:04,239][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:13:04,738][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:13:05,239][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:13:05,738][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:13:06,236][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:13:06,735][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:13:07,233][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:13:07,732][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:13:08,230][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:13:08,729][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:13:09,228][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:13:09,726][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:13:10,225][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:13:10,723][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:13:11,222][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:13:11,720][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:13:12,219][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:13:12,719][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:13:13,220][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:13:13,720][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:13:14,218][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:13:14,716][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:13:15,215][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:13:15,716][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:13:16,215][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:13:16,713][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:13:17,211][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:13:17,709][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:13:18,209][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:13:18,707][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:13:19,205][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:13:19,701][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:13:20,199][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:13:20,698][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:13:21,218][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:13:21,721][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:13:22,222][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:13:22,723][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:13:23,224][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:13:23,724][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:13:24,225][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:13:24,725][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:13:25,225][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:13:25,725][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:13:26,223][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:13:26,721][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:13:27,221][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:13:27,719][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:13:28,218][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:13:28,720][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:13:29,219][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:13:29,718][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:13:30,216][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:13:30,715][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:13:31,214][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:13:31,714][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:13:32,212][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:13:32,710][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:13:33,209][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:13:33,706][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:13:34,205][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:13:34,703][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:13:35,202][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:13:35,702][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:13:36,204][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:13:36,703][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:13:37,202][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:13:37,700][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21486 tokens. [2026-03-25 22:13:38,361][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 22:13:39,136][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:13:39,138][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:13:39,140][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:13:39,942][__main__][INFO] - Iteration 72 took 1m 17s (12.44% Gen, 86.52% Train). Generation: 9s, Training: 1m 6s. Estimated remaining time: 62h 42m 7s. Estimated total time: 64h 21m 5s. Time estimates for 10 more iterations: 12m 52s, 100 more iterations: 2h 8m 42s, 500 more iterations: 10h 43m 30s. [2026-03-25 22:13:39,944][__main__][INFO] - Starting iteration 72. [2026-03-25 22:13:40,346][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:13:40,347][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:13:40,934][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:13:50,181][__main__][INFO] - Number of regex retries in iteration 72: 1 [2026-03-25 22:13:50,181][__main__][INFO] - agents played in iteration 72 are Alice, Bob [2026-03-25 22:13:51,065][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:13:51,109][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:13:51,109][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:13:51,110][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:13:51,722][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:13:52,177][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:13:52,681][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:13:53,181][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:13:53,679][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:13:54,179][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:13:54,679][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:13:55,179][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:13:55,679][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:13:56,177][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:13:56,676][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:13:57,174][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:13:57,673][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:13:58,171][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:13:58,670][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:13:59,167][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:13:59,665][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:14:00,164][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:14:00,664][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:14:01,162][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:14:01,664][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:14:02,163][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:14:02,662][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:14:03,161][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:14:03,660][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:14:04,159][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:14:04,658][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:14:05,157][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:14:05,656][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:14:06,154][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:14:06,653][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:14:07,152][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:14:07,650][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:14:08,147][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:14:08,644][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:14:09,143][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:14:09,640][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:14:10,138][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:14:10,636][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:14:11,134][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:14:11,631][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:14:12,129][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:14:12,627][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:14:13,122][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:14:13,621][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:14:14,120][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:14:14,619][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:14:15,118][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:14:15,616][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:14:16,116][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:14:16,615][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:14:17,114][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:14:17,613][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:14:18,114][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:14:18,615][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:14:19,114][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:14:19,613][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:14:20,113][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:14:20,613][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:14:21,115][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:14:21,614][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:14:22,114][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:14:22,614][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:14:23,113][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:14:23,614][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:14:24,114][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:14:24,615][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:14:25,115][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:14:25,615][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:14:26,115][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:14:26,615][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:14:27,115][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:14:27,614][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:14:28,113][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:14:28,613][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:14:29,111][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:14:29,610][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:14:30,109][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:14:30,607][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:14:31,105][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:14:31,603][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:14:32,103][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:14:32,603][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:14:33,104][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:14:33,603][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:14:34,104][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:14:34,604][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:14:35,103][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:14:35,603][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:14:36,102][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:14:36,602][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:14:37,101][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:14:37,600][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:14:38,099][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:14:38,599][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:14:39,098][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:14:39,599][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:14:40,097][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:14:40,595][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:14:41,093][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:14:41,590][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:14:42,087][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:14:42,581][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:14:43,078][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:14:43,576][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:14:44,075][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:14:44,574][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:14:45,073][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:14:45,575][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:14:46,077][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:14:46,575][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:14:47,075][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:14:47,575][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:14:48,074][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:14:48,575][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:14:49,075][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:14:49,574][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:14:50,073][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:14:50,572][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:14:51,072][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:14:51,572][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:14:52,072][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:14:52,571][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:14:53,069][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:14:53,567][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:14:54,065][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:14:54,564][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:14:55,063][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:14:55,563][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21486 tokens. [2026-03-25 22:14:56,219][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:04 [2026-03-25 22:14:56,956][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:14:56,958][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:14:56,960][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:14:57,607][__main__][INFO] - Iteration 73 took 1m 17s (12.73% Gen, 86.43% Train). Generation: 9s, Training: 1m 6s. Estimated remaining time: 62h 42m 50s. Estimated total time: 64h 23m 5s. Time estimates for 10 more iterations: 12m 52s, 100 more iterations: 2h 8m 46s, 500 more iterations: 10h 43m 50s. [2026-03-25 22:14:57,610][__main__][INFO] - Starting iteration 73. [2026-03-25 22:14:58,010][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:14:58,010][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:14:58,607][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:14:58,613][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:15:06,771][__main__][INFO] - Number of regex retries in iteration 73: 2 [2026-03-25 22:15:06,772][__main__][INFO] - agents played in iteration 73 are Alice, Bob [2026-03-25 22:15:07,655][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:15:07,699][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:15:07,700][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:15:07,700][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:15:08,313][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:15:08,769][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:15:09,274][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:15:09,774][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:15:10,274][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:15:10,773][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:15:11,273][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:15:11,772][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:15:12,271][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:15:12,770][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:15:13,265][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:15:13,764][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:15:14,263][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:15:14,761][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:15:15,259][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:15:15,757][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:15:16,255][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:15:16,753][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:15:17,253][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:15:17,752][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:15:18,253][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:15:18,751][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:15:19,250][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:15:19,748][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:15:20,246][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:15:20,744][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:15:21,242][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:15:21,741][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:15:22,240][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:15:22,739][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:15:23,238][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:15:23,736][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:15:24,234][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:15:24,732][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:15:25,233][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:15:25,732][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:15:26,230][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:15:26,729][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:15:27,227][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:15:27,725][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:15:28,224][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:15:28,723][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:15:29,223][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:15:29,723][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:15:30,222][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:15:30,721][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:15:31,220][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:15:31,720][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:15:32,219][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:15:32,717][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:15:33,215][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:15:33,714][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:15:34,212][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:15:34,711][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:15:35,207][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:15:35,705][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:15:36,203][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:15:36,702][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:15:37,201][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:15:37,700][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:15:38,200][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:15:38,699][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:15:39,198][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:15:39,698][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:15:40,197][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:15:40,696][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:15:41,195][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:15:41,694][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:15:42,194][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:15:42,693][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:15:43,191][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:15:43,691][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:15:44,190][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:15:44,688][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:15:45,186][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:15:45,683][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:15:46,182][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:15:46,680][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:15:47,179][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:15:47,677][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:15:48,175][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:15:48,676][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:15:49,175][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:15:49,674][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:15:50,174][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:15:50,674][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:15:51,173][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:15:51,671][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:15:52,168][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:15:52,663][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:15:53,162][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:15:53,658][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:15:54,157][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:15:54,655][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:15:55,150][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:15:55,646][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:15:56,144][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:15:56,642][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:15:57,141][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:15:57,640][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:15:58,138][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:15:58,636][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:15:59,135][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:15:59,633][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:16:00,152][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:16:00,654][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:16:01,156][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:16:01,657][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:16:02,156][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:16:02,655][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:16:03,154][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:16:03,655][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:16:04,155][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:16:04,654][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:16:05,153][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:16:05,651][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:16:06,149][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:16:06,647][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:16:07,145][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:16:07,642][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:16:08,141][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:16:08,640][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:16:09,139][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:16:09,638][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:16:10,137][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:16:10,636][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:16:11,134][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:16:11,633][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:16:12,133][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21579 tokens. [2026-03-25 22:16:12,787][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:16:13,579][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:16:13,582][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:16:13,583][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:16:14,446][__main__][INFO] - Iteration 74 took 1m 16s (11.46% Gen, 87.41% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 62h 0m 17s. Estimated total time: 63h 41m 50s. Time estimates for 10 more iterations: 12m 44s, 100 more iterations: 2h 7m 23s, 500 more iterations: 10h 36m 58s. [2026-03-25 22:16:14,448][__main__][INFO] - Starting iteration 74. [2026-03-25 22:16:14,848][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:16:14,848][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:16:16,066][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:16:17,865][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:16:22,698][__main__][INFO] - Number of regex retries in iteration 74: 2 [2026-03-25 22:16:22,699][__main__][INFO] - agents played in iteration 74 are Alice, Bob [2026-03-25 22:16:23,577][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:16:23,622][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:16:23,622][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:16:23,623][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:16:24,510][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:16:24,965][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:16:25,467][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:16:25,965][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:16:26,464][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:16:26,963][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:16:27,462][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:16:27,960][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:16:28,460][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:16:28,961][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:16:29,462][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:16:29,961][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:16:30,461][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:16:30,961][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:16:31,461][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:16:31,960][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:16:32,460][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:16:32,958][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:16:33,456][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:16:33,954][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:16:34,449][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:16:34,947][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:16:35,444][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:16:35,942][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:16:36,440][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:16:36,939][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:16:37,437][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:16:37,935][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:16:38,433][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:16:38,930][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:16:39,429][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:16:39,927][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:16:40,424][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:16:40,923][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:16:41,422][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:16:41,921][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:16:42,419][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:16:42,918][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:16:43,417][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:16:43,917][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:16:44,416][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:16:44,914][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:16:45,414][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:16:45,914][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:16:46,413][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:16:46,914][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:16:47,414][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:16:47,914][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:16:48,415][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:16:48,915][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:16:49,413][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:16:49,911][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:16:50,410][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:16:50,908][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:16:51,406][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:16:51,904][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:16:52,401][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:16:52,900][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:16:53,397][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:16:53,895][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:16:54,393][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:16:54,890][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:16:55,388][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:16:55,885][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:16:56,382][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:16:56,881][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:16:57,380][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:16:57,879][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:16:58,378][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:16:58,876][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:16:59,375][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:16:59,875][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:17:00,374][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:17:00,874][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:17:01,373][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:17:01,872][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:17:02,371][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:17:02,869][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:17:03,368][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:17:03,867][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:17:04,366][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:17:04,864][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:17:05,360][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:17:05,858][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:17:06,355][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:17:06,852][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:17:07,348][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:17:07,845][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:17:08,342][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:17:08,839][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:17:09,338][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:17:09,836][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:17:10,332][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:17:10,830][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:17:11,328][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:17:11,825][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:17:12,323][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:17:12,822][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:17:13,321][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:17:13,820][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:17:14,318][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:17:14,816][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:17:15,315][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:17:15,814][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:17:16,314][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:17:16,814][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:17:17,314][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:17:17,809][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:17:18,311][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:17:18,807][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:17:19,302][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:17:19,799][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:17:20,296][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:17:20,795][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:17:21,293][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:17:21,792][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:17:22,290][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:17:22,788][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:17:23,285][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:17:23,782][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:17:24,280][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:17:24,778][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:17:25,277][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:17:25,776][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:17:26,275][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:17:26,774][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:17:27,271][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:17:27,768][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:17:28,266][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21543 tokens. [2026-03-25 22:17:28,902][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:17:29,643][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:17:29,645][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:17:29,647][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:17:30,353][__main__][INFO] - Iteration 75 took 1m 15s (10.40% Gen, 88.67% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 61h 12m 28s. Estimated total time: 62h 55m 16s. Time estimates for 10 more iterations: 12m 35s, 100 more iterations: 2h 5m 50s, 500 more iterations: 10h 29m 12s. [2026-03-25 22:17:30,355][__main__][INFO] - Starting iteration 75. [2026-03-25 22:17:30,755][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:17:30,755][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:17:39,425][__main__][INFO] - Number of regex retries in iteration 75: 0 [2026-03-25 22:17:39,426][__main__][INFO] - agents played in iteration 75 are Alice, Bob [2026-03-25 22:17:40,321][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:17:40,365][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:17:40,366][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:17:40,366][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:17:40,972][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:17:41,428][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:17:41,931][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:17:42,429][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:17:42,928][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:17:43,427][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:17:43,925][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:17:44,423][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:17:44,922][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:17:45,422][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:17:45,923][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:17:46,421][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:17:46,920][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:17:47,418][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:17:47,916][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:17:48,416][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:17:48,915][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:17:49,413][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:17:49,911][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:17:50,409][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:17:50,907][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:17:51,406][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:17:51,904][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:17:52,402][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:17:52,900][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:17:53,401][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:17:53,900][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:17:54,400][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:17:54,899][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:17:55,400][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:17:55,899][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:17:56,397][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:17:56,897][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:17:57,395][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:17:57,889][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:17:58,387][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:17:58,884][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:17:59,381][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:17:59,880][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:18:00,379][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:18:00,877][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:18:01,376][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:18:01,874][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:18:02,371][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:18:02,871][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:18:03,369][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:18:03,867][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:18:04,366][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:18:04,863][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:18:05,364][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:18:05,863][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:18:06,362][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:18:06,862][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:18:07,362][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:18:07,861][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:18:08,361][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:18:08,861][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:18:09,359][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:18:09,857][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:18:10,355][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:18:10,855][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:18:11,354][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:18:11,853][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:18:12,352][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:18:12,850][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:18:13,348][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:18:13,846][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:18:14,344][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:18:14,843][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:18:15,341][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:18:15,840][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:18:16,339][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:18:16,837][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:18:17,335][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:18:17,833][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:18:18,332][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:18:18,832][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:18:19,330][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:18:19,828][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:18:20,326][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:18:20,823][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:18:21,321][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:18:21,820][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:18:22,318][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:18:22,817][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:18:23,315][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:18:23,814][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:18:24,313][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:18:24,811][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:18:25,310][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:18:25,808][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:18:26,306][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:18:26,805][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:18:27,303][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:18:27,803][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:18:28,301][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:18:28,800][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:18:29,298][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:18:29,797][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:18:30,297][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:18:30,796][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:18:31,296][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:18:31,795][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:18:32,295][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:18:32,794][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:18:33,293][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:18:33,791][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:18:34,289][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:18:34,787][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:18:35,285][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:18:35,783][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:18:36,282][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:18:36,780][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:18:37,279][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:18:37,779][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:18:38,279][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:18:38,778][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:18:39,278][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:18:39,777][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:18:40,277][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:18:40,776][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:18:41,274][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:18:41,772][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:18:42,269][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:18:42,768][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:18:43,266][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:18:43,764][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:18:44,263][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:18:44,761][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21595 tokens. [2026-03-25 22:18:45,404][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 22:18:46,138][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:18:46,141][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:18:46,142][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:18:46,864][__main__][INFO] - Iteration 76 took 1m 16s (11.39% Gen, 87.66% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 41m 28s. Estimated total time: 63h 25m 33s. Time estimates for 10 more iterations: 12m 41s, 100 more iterations: 2h 6m 51s, 500 more iterations: 10h 34m 15s. [2026-03-25 22:18:46,867][__main__][INFO] - Starting iteration 76. [2026-03-25 22:18:47,268][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:18:47,268][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:18:47,909][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:18:56,852][__main__][INFO] - Number of regex retries in iteration 76: 1 [2026-03-25 22:18:56,853][__main__][INFO] - agents played in iteration 76 are Alice, Bob [2026-03-25 22:18:57,737][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:18:57,781][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:18:57,782][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:18:57,782][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:18:58,394][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:18:58,850][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:18:59,351][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:18:59,849][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:19:00,348][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:19:00,846][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:19:01,344][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:19:01,843][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:19:02,343][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:19:02,841][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:19:03,340][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:19:03,838][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:19:04,337][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:19:04,835][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:19:05,333][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:19:05,831][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:19:06,331][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:19:06,829][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:19:07,327][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:19:07,825][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:19:08,324][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:19:08,823][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:19:09,321][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:19:09,820][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:19:10,319][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:19:10,815][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:19:11,313][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:19:11,809][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:19:12,305][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:19:12,802][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:19:13,300][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:19:13,797][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:19:14,294][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:19:14,795][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:19:15,294][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:19:15,797][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:19:16,296][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:19:16,796][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:19:17,296][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:19:17,795][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:19:18,296][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:19:18,794][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:19:19,293][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:19:19,791][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:19:20,288][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:19:20,786][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:19:21,282][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:19:21,782][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:19:22,277][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:19:22,777][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:19:23,276][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:19:23,775][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:19:24,274][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:19:24,773][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:19:25,276][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:19:25,775][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:19:26,274][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:19:26,772][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:19:27,274][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:19:27,772][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:19:28,271][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:19:28,769][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:19:29,268][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:19:29,766][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:19:30,264][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:19:30,762][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:19:31,260][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:19:31,759][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:19:32,257][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:19:32,756][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:19:33,255][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:19:33,754][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:19:34,253][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:19:34,752][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:19:35,250][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:19:35,750][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:19:36,249][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:19:36,747][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:19:37,246][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:19:37,744][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:19:38,243][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:19:38,742][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:19:39,242][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:19:39,741][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:19:40,239][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:19:40,738][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:19:41,236][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:19:41,734][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:19:42,232][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:19:42,728][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:19:43,224][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:19:43,721][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:19:44,218][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:19:44,715][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:19:45,211][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:19:45,708][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:19:46,205][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:19:46,701][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:19:47,201][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:19:47,699][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:19:48,199][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:19:48,699][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:19:49,197][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:19:49,693][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:19:50,192][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:19:50,690][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:19:51,189][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:19:51,686][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:19:52,184][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:19:52,682][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:19:53,180][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:19:53,678][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:19:54,177][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:19:54,675][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:19:55,175][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:19:55,674][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:19:56,171][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:19:56,669][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:19:57,166][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:19:57,665][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:19:58,164][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:19:58,664][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:19:59,163][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:19:59,663][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:20:00,163][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:20:00,661][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:20:01,160][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:20:01,661][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:20:02,161][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21515 tokens. [2026-03-25 22:20:02,802][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.27%, ΔTime: 00:01:04 [2026-03-25 22:20:03,544][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:20:03,547][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:20:03,548][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:20:04,201][__main__][INFO] - Iteration 77 took 1m 16s (12.46% Gen, 86.69% Train). Generation: 9s, Training: 1m 6s. Estimated remaining time: 62h 21m 19s. Estimated total time: 64h 6m 41s. Time estimates for 10 more iterations: 12m 49s, 100 more iterations: 2h 8m 13s, 500 more iterations: 10h 41m 6s. [2026-03-25 22:20:04,203][__main__][INFO] - Starting iteration 77. [2026-03-25 22:20:04,605][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:20:04,605][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:20:12,421][__main__][INFO] - Number of regex retries in iteration 77: 0 [2026-03-25 22:20:12,422][__main__][INFO] - agents played in iteration 77 are Alice, Bob [2026-03-25 22:20:13,308][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:20:13,353][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:20:13,354][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:20:13,354][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:20:14,241][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:20:14,696][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:20:15,199][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:20:15,697][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:20:16,197][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:20:16,696][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:20:17,194][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:20:17,695][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:20:18,195][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:20:18,695][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:20:19,195][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:20:19,694][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:20:20,193][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:20:20,692][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:20:21,190][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:20:21,690][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:20:22,189][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:20:22,688][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:20:23,186][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:20:23,683][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:20:24,181][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:20:24,680][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:20:25,179][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:20:25,677][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:20:26,177][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:20:26,675][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:20:27,173][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:20:27,671][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:20:28,168][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:20:28,665][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:20:29,163][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:20:29,661][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:20:30,159][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:20:30,660][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:20:31,159][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:20:31,658][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:20:32,156][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:20:32,654][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:20:33,152][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:20:33,651][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:20:34,150][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:20:34,648][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:20:35,146][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:20:35,644][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:20:36,142][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:20:36,640][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:20:37,138][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:20:37,635][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:20:38,134][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:20:38,634][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:20:39,133][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:20:39,633][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:20:40,132][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:20:40,630][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:20:41,128][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:20:41,626][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:20:42,125][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:20:42,624][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:20:43,123][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:20:43,622][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:20:44,121][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:20:44,620][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:20:45,119][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:20:45,617][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:20:46,114][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:20:46,615][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:20:47,114][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:20:47,613][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:20:48,111][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:20:48,610][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:20:49,107][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:20:49,603][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:20:50,102][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:20:50,602][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:20:51,101][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:20:51,600][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:20:52,098][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:20:52,596][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:20:53,095][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:20:53,593][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:20:54,092][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:20:54,590][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:20:55,089][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:20:55,588][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:20:56,086][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:20:56,583][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:20:57,082][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:20:57,580][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:20:58,078][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:20:58,576][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:20:59,075][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:20:59,572][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:21:00,069][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:21:00,566][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:21:01,063][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:21:01,560][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:21:02,058][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:21:02,556][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:21:03,057][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:21:03,556][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:21:04,056][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:21:04,555][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:21:05,054][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:21:05,552][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:21:06,050][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:21:06,547][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:21:07,044][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:21:07,541][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:21:08,039][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:21:08,537][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:21:09,035][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:21:09,533][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:21:10,030][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:21:10,527][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:21:11,026][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:21:11,525][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:21:12,025][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:21:12,524][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:21:13,023][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:21:13,522][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:21:14,021][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:21:14,521][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:21:15,020][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:21:15,520][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:21:16,019][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:21:16,517][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:21:17,017][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:21:17,515][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:21:18,014][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21635 tokens. [2026-03-25 22:21:18,649][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 22:21:19,357][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:21:19,360][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:21:19,361][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:21:20,049][__main__][INFO] - Iteration 78 took 1m 15s (10.36% Gen, 88.73% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 61h 5m 36s. Estimated total time: 62h 52m 14s. Time estimates for 10 more iterations: 12m 34s, 100 more iterations: 2h 5m 44s, 500 more iterations: 10h 28m 42s. [2026-03-25 22:21:20,051][__main__][INFO] - Starting iteration 78. [2026-03-25 22:21:20,457][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:21:20,457][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:21:29,433][__main__][INFO] - Number of regex retries in iteration 78: 0 [2026-03-25 22:21:29,434][__main__][INFO] - agents played in iteration 78 are Alice, Bob [2026-03-25 22:21:30,317][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:21:30,362][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:21:30,363][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:21:30,363][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:21:31,253][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:21:31,709][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:21:32,212][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:21:32,710][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:21:33,206][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:21:33,706][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:21:34,204][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:21:34,704][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:21:35,203][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:21:35,702][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:21:36,200][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:21:36,699][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:21:37,195][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:21:37,689][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:21:38,187][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:21:38,685][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:21:39,180][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:21:39,679][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:21:40,177][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:21:40,674][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:21:41,171][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:21:41,667][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:21:42,165][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:21:42,663][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:21:43,161][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:21:43,661][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:21:44,163][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:21:44,662][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:21:45,162][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:21:45,660][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:21:46,159][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:21:46,659][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:21:47,158][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:21:47,657][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:21:48,156][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:21:48,656][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:21:49,153][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:21:49,651][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:21:50,149][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:21:50,646][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:21:51,144][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:21:51,644][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:21:52,143][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:21:52,642][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:21:53,142][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:21:53,640][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:21:54,139][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:21:54,638][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:21:55,138][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:21:55,637][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:21:56,136][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:21:56,635][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:21:57,133][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:21:57,631][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:21:58,130][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:21:58,628][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:21:59,127][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:21:59,625][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:22:00,123][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:22:00,622][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:22:01,121][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:22:01,619][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:22:02,118][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:22:02,618][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:22:03,117][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:22:03,616][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:22:04,115][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:22:04,609][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:22:05,105][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:22:05,600][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:22:06,099][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:22:06,598][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:22:07,093][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:22:07,591][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:22:08,089][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:22:08,586][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:22:09,084][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:22:09,582][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:22:10,079][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:22:10,577][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:22:11,075][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:22:11,572][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:22:12,069][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:22:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:22:13,064][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:22:13,562][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:22:14,060][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:22:14,557][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:22:15,055][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:22:15,555][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:22:16,055][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:22:16,556][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:22:17,056][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:22:17,555][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:22:18,055][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:22:18,555][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:22:19,053][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:22:19,551][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:22:20,049][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:22:20,546][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:22:21,044][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:22:21,542][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:22:22,041][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:22:22,539][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:22:23,037][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:22:23,536][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:22:24,035][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:22:24,534][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:22:25,034][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:22:25,533][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:22:26,033][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:22:26,531][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:22:27,030][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:22:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:22:28,028][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:22:28,528][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:22:29,027][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:22:29,527][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:22:30,026][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:22:30,526][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:22:31,025][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:22:31,522][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:22:32,018][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:22:32,514][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:22:33,009][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:22:33,503][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:22:33,998][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:22:34,492][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:22:34,986][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21375 tokens. [2026-03-25 22:22:35,625][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.16%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:22:36,364][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:22:36,367][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:22:36,368][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:22:37,014][__main__][INFO] - Iteration 79 took 1m 16s (11.73% Gen, 87.43% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 59m 58s. Estimated total time: 63h 47m 53s. Time estimates for 10 more iterations: 12m 45s, 100 more iterations: 2h 7m 35s, 500 more iterations: 10h 37m 58s. [2026-03-25 22:22:37,016][__main__][INFO] - Starting iteration 79. [2026-03-25 22:22:37,417][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:22:37,418][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:22:46,430][__main__][INFO] - Number of regex retries in iteration 79: 0 [2026-03-25 22:22:46,431][__main__][INFO] - agents played in iteration 79 are Alice, Bob [2026-03-25 22:22:47,309][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:22:47,354][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:22:47,354][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:22:47,355][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:22:47,959][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:22:48,417][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:22:48,920][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:22:49,419][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:22:49,918][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:22:50,418][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:22:50,917][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:22:51,417][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:22:51,917][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:22:52,416][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:22:52,914][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:22:53,413][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:22:53,911][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:22:54,409][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:22:54,907][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:22:55,406][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:22:55,904][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:22:56,406][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:22:56,905][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:22:57,403][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:22:57,902][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:22:58,401][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:22:58,900][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:22:59,399][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:22:59,898][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:23:00,394][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:23:00,891][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:23:01,388][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:23:01,884][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:23:02,381][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:23:02,878][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:23:03,376][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:23:03,872][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:23:04,371][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:23:04,870][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:23:05,369][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:23:05,868][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:23:06,366][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:23:06,865][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:23:07,364][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:23:07,864][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:23:08,363][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:23:08,863][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:23:09,362][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:23:09,860][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:23:10,357][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:23:10,855][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:23:11,353][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:23:11,851][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:23:12,348][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:23:12,846][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:23:13,343][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:23:13,841][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:23:14,339][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:23:14,836][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:23:15,333][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:23:15,830][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:23:16,328][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:23:16,827][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:23:17,325][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:23:17,824][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:23:18,325][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:23:18,823][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:23:19,322][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:23:19,821][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:23:20,320][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:23:20,819][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:23:21,317][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:23:21,817][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:23:22,317][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:23:22,816][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:23:23,314][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:23:23,813][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:23:24,311][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:23:24,809][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:23:25,308][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:23:25,806][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:23:26,305][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:23:26,802][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:23:27,298][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:23:27,796][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:23:28,294][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:23:28,794][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:23:29,293][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:23:29,791][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:23:30,289][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:23:30,788][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:23:31,285][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:23:31,783][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:23:32,278][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:23:32,775][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:23:33,271][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:23:33,767][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:23:34,262][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:23:34,759][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:23:35,256][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:23:35,753][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:23:36,251][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:23:36,750][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:23:37,248][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:23:37,748][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:23:38,247][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:23:38,746][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:23:39,245][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:23:39,744][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:23:40,242][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:23:40,739][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:23:41,236][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:23:41,734][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:23:42,233][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:23:42,732][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:23:43,230][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:23:43,729][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:23:44,227][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:23:44,725][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:23:45,223][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:23:45,724][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:23:46,223][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:23:46,722][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:23:47,220][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:23:47,718][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:23:48,220][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:23:48,716][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:23:49,215][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:23:49,713][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:23:50,208][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:23:50,707][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:23:51,205][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:23:51,705][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21389 tokens. [2026-03-25 22:23:52,330][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 22:23:53,061][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:23:53,064][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:23:53,065][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:23:53,711][__main__][INFO] - Iteration 80 took 1m 16s (11.81% Gen, 87.34% Train). Generation: 9s, Training: 1m 6s. Estimated remaining time: 61h 45m 31s. Estimated total time: 63h 34m 43s. Time estimates for 10 more iterations: 12m 42s, 100 more iterations: 2h 7m 9s, 500 more iterations: 10h 35m 47s. [2026-03-25 22:23:53,713][__main__][INFO] - Starting iteration 80. [2026-03-25 22:23:54,121][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:23:54,121][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:23:54,706][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:24:02,862][__main__][INFO] - Number of regex retries in iteration 80: 1 [2026-03-25 22:24:02,862][__main__][INFO] - agents played in iteration 80 are Alice, Bob [2026-03-25 22:24:03,740][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:24:03,784][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:24:03,785][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:24:03,785][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:24:04,395][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:24:04,852][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:24:05,355][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:24:05,854][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:24:06,353][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:24:06,852][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:24:07,350][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:24:07,848][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:24:08,347][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:24:08,845][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:24:09,342][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:24:09,839][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:24:10,337][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:24:10,835][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:24:11,333][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:24:11,830][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:24:12,327][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:24:12,826][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:24:13,324][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:24:13,822][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:24:14,321][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:24:14,820][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:24:15,318][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:24:15,817][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:24:16,315][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:24:16,816][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:24:17,317][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:24:17,817][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:24:18,319][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:24:18,819][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:24:19,321][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:24:19,821][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:24:20,321][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:24:20,820][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:24:21,319][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:24:21,818][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:24:22,318][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:24:22,818][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:24:23,318][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:24:23,818][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:24:24,318][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:24:24,817][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:24:25,317][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:24:25,815][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:24:26,314][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:24:26,812][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:24:27,312][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:24:27,810][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:24:28,308][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:24:28,806][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:24:29,303][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:24:29,801][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:24:30,298][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:24:30,796][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:24:31,294][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:24:31,788][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:24:32,287][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:24:32,785][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:24:33,281][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:24:33,778][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:24:34,277][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:24:34,769][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:24:35,264][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:24:35,762][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:24:36,260][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:24:36,758][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:24:37,258][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:24:37,758][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:24:38,257][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:24:38,756][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:24:39,254][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:24:39,752][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:24:40,251][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:24:40,748][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:24:41,246][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:24:41,745][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:24:42,243][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:24:42,741][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:24:43,240][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:24:43,737][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:24:44,235][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:24:44,732][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:24:45,231][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:24:45,729][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:24:46,228][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:24:46,727][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:24:47,224][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:24:47,722][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:24:48,223][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:24:48,723][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:24:49,222][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:24:49,721][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:24:50,220][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:24:50,719][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:24:51,218][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:24:51,717][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:24:52,216][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:24:52,715][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:24:53,214][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:24:53,712][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:24:54,211][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:24:54,710][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:24:55,209][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:24:55,707][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:24:56,206][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:24:56,705][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:24:57,203][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:24:57,703][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:24:58,201][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:24:58,699][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:24:59,197][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:24:59,695][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:25:00,193][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:25:00,690][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:25:01,186][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:25:01,684][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:25:02,182][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:25:02,679][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:25:03,176][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:25:03,674][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:25:04,170][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:25:04,668][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:25:05,165][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:25:05,663][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:25:06,161][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:25:06,660][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:25:07,158][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:25:07,657][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:25:08,155][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21602 tokens. [2026-03-25 22:25:08,789][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.30%, ΔTime: 00:01:04 [2026-03-25 22:25:09,532][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:25:09,534][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:25:09,536][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:25:10,182][__main__][INFO] - Iteration 81 took 1m 16s (11.49% Gen, 87.66% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 32m 38s. Estimated total time: 63h 23m 6s. Time estimates for 10 more iterations: 12m 40s, 100 more iterations: 2h 6m 46s, 500 more iterations: 10h 33m 51s. [2026-03-25 22:25:10,184][__main__][INFO] - Starting iteration 81. [2026-03-25 22:25:10,585][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:25:10,586][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:25:11,172][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:25:19,450][__main__][INFO] - Number of regex retries in iteration 81: 1 [2026-03-25 22:25:19,451][__main__][INFO] - agents played in iteration 81 are Alice, Bob [2026-03-25 22:25:20,336][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:25:20,381][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:25:20,381][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:25:20,382][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:25:20,991][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:25:21,446][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:25:21,950][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:25:22,449][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:25:22,946][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:25:23,448][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:25:23,945][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:25:24,446][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:25:24,943][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:25:25,445][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:25:25,946][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:25:26,446][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:25:26,946][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:25:27,446][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:25:27,946][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:25:28,446][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:25:28,946][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:25:29,445][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:25:29,943][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:25:30,443][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:25:30,942][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:25:31,440][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:25:31,939][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:25:32,437][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:25:32,936][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:25:33,436][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:25:33,935][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:25:34,435][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:25:34,935][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:25:35,436][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:25:35,935][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:25:36,435][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:25:36,935][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:25:37,434][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:25:37,933][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:25:38,434][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:25:38,934][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:25:39,433][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:25:39,933][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:25:40,433][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:25:40,932][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:25:41,430][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:25:41,928][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:25:42,427][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:25:42,927][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:25:43,426][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:25:43,926][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:25:44,426][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:25:44,924][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:25:45,423][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:25:45,923][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:25:46,421][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:25:46,920][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:25:47,418][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:25:47,916][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:25:48,416][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:25:48,914][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:25:49,412][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:25:49,910][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:25:50,408][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:25:50,907][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:25:51,405][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:25:51,903][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:25:52,402][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:25:52,901][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:25:53,400][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:25:53,899][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:25:54,398][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:25:54,898][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:25:55,398][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:25:55,897][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:25:56,396][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:25:56,894][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:25:57,393][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:25:57,893][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:25:58,392][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:25:58,893][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:25:59,393][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:25:59,893][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:26:00,393][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:26:00,892][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:26:01,389][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:26:01,888][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:26:02,386][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:26:02,885][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:26:03,383][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:26:03,882][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:26:04,381][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:26:04,878][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:26:05,377][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:26:05,876][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:26:06,376][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:26:06,874][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:26:07,371][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:26:07,870][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:26:08,369][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:26:08,867][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:26:09,366][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:26:09,866][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:26:10,366][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:26:10,866][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:26:11,366][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:26:11,866][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:26:12,366][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:26:12,866][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:26:13,364][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:26:13,863][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:26:14,362][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:26:14,861][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:26:15,359][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:26:15,858][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:26:16,357][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:26:16,855][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:26:17,352][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:26:17,848][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:26:18,346][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:26:18,845][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:26:19,341][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:26:19,836][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:26:20,332][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:26:20,828][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:26:21,326][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:26:21,824][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:26:22,323][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:26:22,822][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:26:23,321][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:26:23,819][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:26:24,318][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:26:24,816][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21511 tokens. [2026-03-25 22:26:25,453][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 22:26:26,198][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:26:26,200][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:26:26,202][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:26:27,045][__main__][INFO] - Iteration 82 took 1m 16s (11.59% Gen, 87.30% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 51m 17s. Estimated total time: 63h 43m 2s. Time estimates for 10 more iterations: 12m 44s, 100 more iterations: 2h 7m 26s, 500 more iterations: 10h 37m 10s. [2026-03-25 22:26:27,048][__main__][INFO] - Starting iteration 82. [2026-03-25 22:26:27,452][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:26:27,452][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:26:36,044][__main__][INFO] - Number of regex retries in iteration 82: 0 [2026-03-25 22:26:36,045][__main__][INFO] - agents played in iteration 82 are Alice, Bob [2026-03-25 22:26:36,923][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:26:36,968][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:26:36,968][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:26:36,969][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:26:37,593][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:26:38,050][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:26:38,554][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:26:39,053][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:26:39,552][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:26:40,052][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:26:40,551][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:26:41,050][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:26:41,549][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:26:42,048][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:26:42,546][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:26:43,045][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:26:43,543][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:26:44,042][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:26:44,541][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:26:45,038][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:26:45,539][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:26:46,040][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:26:46,538][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:26:47,038][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:26:47,538][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:26:48,037][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:26:48,539][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:26:49,037][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:26:49,535][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:26:50,033][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:26:50,531][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:26:51,030][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:26:51,528][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:26:52,027][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:26:52,525][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:26:53,024][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:26:53,523][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:26:54,022][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:26:54,520][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:26:55,021][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:26:55,520][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:26:56,020][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:26:56,519][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:26:57,017][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:26:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:26:58,013][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:26:58,508][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:26:59,005][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:26:59,501][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:26:59,998][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:27:00,491][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:27:00,987][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:27:01,486][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:27:01,984][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:27:02,481][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:27:02,978][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:27:03,475][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:27:03,973][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:27:04,469][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:27:04,966][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:27:05,464][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:27:05,962][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:27:06,460][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:27:06,957][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:27:07,455][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:27:07,952][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:27:08,451][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:27:08,949][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:27:09,447][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:27:09,948][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:27:10,447][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:27:10,947][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:27:11,447][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:27:11,946][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:27:12,445][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:27:12,945][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:27:13,444][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:27:13,943][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:27:14,442][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:27:14,940][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:27:15,438][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:27:15,936][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:27:16,435][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:27:16,933][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:27:17,432][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:27:17,932][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:27:18,432][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:27:18,931][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:27:19,429][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:27:19,927][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:27:20,427][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:27:20,926][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:27:21,425][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:27:21,924][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:27:22,423][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:27:22,922][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:27:23,422][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:27:23,921][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:27:24,420][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:27:24,920][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:27:25,420][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:27:25,920][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:27:26,419][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:27:26,918][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:27:27,417][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:27:27,917][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:27:28,417][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:27:28,916][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:27:29,414][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:27:29,912][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:27:30,411][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:27:30,909][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:27:31,407][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:27:31,906][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:27:32,405][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:27:32,903][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:27:33,402][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:27:33,900][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:27:34,399][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:27:34,897][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:27:35,396][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:27:35,895][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:27:36,393][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:27:36,890][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:27:37,389][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:27:37,887][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:27:38,385][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:27:38,884][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:27:39,382][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:27:39,881][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:27:40,380][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:27:40,879][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:27:41,377][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21545 tokens. [2026-03-25 22:27:42,023][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 22:27:42,764][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:27:42,767][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:27:42,768][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:27:43,569][__main__][INFO] - Iteration 83 took 1m 16s (11.29% Gen, 87.66% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 32m 52s. Estimated total time: 63h 25m 54s. Time estimates for 10 more iterations: 12m 41s, 100 more iterations: 2h 6m 51s, 500 more iterations: 10h 34m 19s. [2026-03-25 22:27:43,571][__main__][INFO] - Starting iteration 83. [2026-03-25 22:27:43,970][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:27:43,970][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:27:51,962][__main__][INFO] - Number of regex retries in iteration 83: 0 [2026-03-25 22:27:51,963][__main__][INFO] - agents played in iteration 83 are Alice, Bob [2026-03-25 22:27:52,841][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:27:52,885][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:27:52,886][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:27:52,886][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:27:53,506][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:27:53,963][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:27:54,466][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:27:54,966][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:27:55,466][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:27:55,965][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:27:56,464][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:27:56,964][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:27:57,464][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:27:57,965][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:27:58,464][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:27:58,964][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:27:59,463][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:27:59,963][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:28:00,463][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:28:00,962][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:28:01,462][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:28:01,961][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:28:02,461][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:28:02,959][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:28:03,457][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:28:03,955][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:28:04,454][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:28:04,953][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:28:05,452][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:28:05,950][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:28:06,449][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:28:06,948][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:28:07,446][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:28:07,944][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:28:08,443][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:28:08,942][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:28:09,441][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:28:09,940][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:28:10,438][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:28:10,935][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:28:11,430][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:28:11,927][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:28:12,424][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:28:12,922][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:28:13,420][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:28:13,920][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:28:14,419][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:28:14,919][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:28:15,418][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:28:15,917][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:28:16,416][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:28:16,917][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:28:17,417][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:28:17,919][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:28:18,421][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:28:18,923][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:28:19,424][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:28:19,924][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:28:20,424][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:28:20,925][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:28:21,425][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:28:21,924][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:28:22,423][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:28:22,921][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:28:23,420][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:28:23,920][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:28:24,419][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:28:24,918][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:28:25,418][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:28:25,916][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:28:26,414][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:28:26,911][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:28:27,408][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:28:27,906][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:28:28,404][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:28:28,901][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:28:29,399][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:28:29,898][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:28:30,397][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:28:30,895][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:28:31,393][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:28:31,891][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:28:32,389][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:28:32,888][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:28:33,387][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:28:33,881][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:28:34,381][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:28:34,879][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:28:35,377][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:28:35,874][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:28:36,373][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:28:36,871][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:28:37,369][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:28:37,867][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:28:38,365][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:28:38,863][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:28:39,361][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:28:39,855][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:28:40,354][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:28:40,852][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:28:41,349][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:28:41,847][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:28:42,344][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:28:42,842][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:28:43,339][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:28:43,836][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:28:44,333][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:28:44,830][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:28:45,328][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:28:45,828][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:28:46,328][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:28:46,827][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:28:47,326][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:28:47,825][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:28:48,324][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:28:48,823][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:28:49,322][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:28:49,823][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:28:50,323][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:28:50,823][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:28:51,323][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:28:51,823][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:28:52,323][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:28:52,823][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:28:53,324][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:28:53,822][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:28:54,321][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:28:54,820][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:28:55,319][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:28:55,818][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:28:56,318][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:28:56,817][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:28:57,316][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21694 tokens. [2026-03-25 22:28:57,945][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 22:28:58,713][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:28:58,715][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:28:58,717][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:28:59,475][__main__][INFO] - Iteration 84 took 1m 15s (10.59% Gen, 88.41% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 61h 1m 1s. Estimated total time: 62h 55m 19s. Time estimates for 10 more iterations: 12m 35s, 100 more iterations: 2h 5m 50s, 500 more iterations: 10h 29m 13s. [2026-03-25 22:28:59,478][__main__][INFO] - Starting iteration 84. [2026-03-25 22:28:59,878][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:28:59,879][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:29:00,463][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:29:08,288][__main__][INFO] - Number of regex retries in iteration 84: 1 [2026-03-25 22:29:08,568][__main__][INFO] - agents played in iteration 84 are Alice, Bob [2026-03-25 22:29:09,445][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:29:09,489][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:29:09,490][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:29:09,490][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:29:10,094][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:29:10,549][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:29:11,051][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:29:11,549][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:29:12,047][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:29:12,545][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:29:13,044][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:29:13,542][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:29:14,040][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:29:14,541][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:29:15,040][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:29:15,535][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:29:16,033][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:29:16,531][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:29:17,030][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:29:17,527][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:29:18,026][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:29:18,527][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:29:19,026][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:29:19,525][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:29:20,024][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:29:20,523][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:29:21,023][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:29:21,522][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:29:22,021][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:29:22,520][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:29:23,018][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:29:23,517][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:29:24,015][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:29:24,512][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:29:25,010][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:29:25,508][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:29:26,006][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:29:26,508][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:29:27,008][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:29:27,507][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:29:28,006][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:29:28,505][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:29:29,004][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:29:29,503][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:29:30,002][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:29:30,500][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:29:30,998][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:29:31,495][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:29:31,993][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:29:32,490][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:29:32,987][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:29:33,484][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:29:33,982][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:29:34,481][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:29:34,980][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:29:35,478][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:29:35,977][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:29:36,474][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:29:36,973][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:29:37,471][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:29:37,970][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:29:38,468][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:29:38,965][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:29:39,465][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:29:39,962][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:29:40,460][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:29:40,958][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:29:41,455][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:29:41,952][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:29:42,450][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:29:42,948][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:29:43,446][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:29:43,944][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:29:44,442][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:29:44,940][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:29:45,439][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:29:45,936][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:29:46,434][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:29:46,933][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:29:47,432][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:29:47,931][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:29:48,432][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:29:48,931][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:29:49,430][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:29:49,929][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:29:50,427][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:29:50,926][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:29:51,425][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:29:51,924][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:29:52,423][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:29:52,922][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:29:53,421][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:29:53,921][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:29:54,420][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:29:54,921][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:29:55,421][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:29:55,920][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:29:56,420][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:29:56,918][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:29:57,417][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:29:57,916][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:29:58,416][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:29:58,916][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:29:59,415][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:29:59,915][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:30:00,414][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:30:00,913][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:30:01,411][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:30:01,908][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:30:02,406][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:30:02,905][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:30:03,403][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:30:03,901][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:30:04,399][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:30:04,897][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:30:05,394][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:30:05,890][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:30:06,387][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:30:06,885][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:30:07,384][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:30:07,882][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:30:08,380][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:30:08,877][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:30:09,374][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:30:09,873][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:30:10,368][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:30:10,863][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:30:11,361][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:30:11,860][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:30:12,358][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:30:12,856][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:30:13,353][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:30:13,851][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21635 tokens. [2026-03-25 22:30:14,487][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:30:15,225][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:30:15,227][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:30:15,228][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:30:15,877][__main__][INFO] - Iteration 85 took 1m 15s (11.43% Gen, 87.71% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 24m 22s. Estimated total time: 63h 19m 56s. Time estimates for 10 more iterations: 12m 39s, 100 more iterations: 2h 6m 39s, 500 more iterations: 10h 33m 19s. [2026-03-25 22:30:15,879][__main__][INFO] - Starting iteration 85. [2026-03-25 22:30:16,284][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:30:16,285][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:30:16,877][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:30:19,496][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:30:23,409][__main__][INFO] - Number of regex retries in iteration 85: 2 [2026-03-25 22:30:23,410][__main__][INFO] - agents played in iteration 85 are Alice, Bob [2026-03-25 22:30:24,290][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:30:24,334][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:30:24,335][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:30:24,336][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:30:24,942][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:30:25,396][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:30:25,899][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:30:26,398][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:30:26,896][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:30:27,395][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:30:27,893][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:30:28,391][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:30:28,890][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:30:29,390][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:30:29,889][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:30:30,388][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:30:30,887][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:30:31,386][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:30:31,886][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:30:32,385][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:30:32,884][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:30:33,381][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:30:33,878][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:30:34,376][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:30:34,873][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:30:35,370][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:30:35,868][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:30:36,365][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:30:36,863][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:30:37,362][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:30:37,859][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:30:38,356][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:30:38,854][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:30:39,350][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:30:39,849][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:30:40,347][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:30:40,845][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:30:41,343][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:30:41,841][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:30:42,339][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:30:42,837][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:30:43,334][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:30:43,831][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:30:44,328][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:30:44,826][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:30:45,324][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:30:45,822][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:30:46,320][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:30:46,819][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:30:47,316][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:30:47,814][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:30:48,313][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:30:48,811][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:30:49,309][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:30:49,807][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:30:50,305][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:30:50,803][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:30:51,300][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:30:51,796][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:30:52,294][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:30:52,790][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:30:53,285][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:30:53,779][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:30:54,274][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:30:54,769][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:30:55,266][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:30:55,762][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:30:56,258][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:30:56,754][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:30:57,253][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:30:57,752][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:30:58,252][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:30:58,752][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:30:59,251][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:30:59,750][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:31:00,248][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:31:00,746][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:31:01,245][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:31:01,745][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:31:02,243][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:31:02,742][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:31:03,242][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:31:03,741][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:31:04,241][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:31:04,740][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:31:05,237][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:31:05,735][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:31:06,231][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:31:06,729][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:31:07,226][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:31:07,724][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:31:08,222][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:31:08,720][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:31:09,219][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:31:09,716][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:31:10,214][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:31:10,712][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:31:11,210][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:31:11,708][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:31:12,207][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:31:12,706][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:31:13,204][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:31:13,702][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:31:14,199][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:31:14,697][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:31:15,191][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:31:15,693][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:31:16,190][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:31:16,686][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:31:17,184][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:31:17,682][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:31:18,180][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:31:18,679][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:31:19,176][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:31:19,673][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:31:20,170][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:31:20,668][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:31:21,165][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:31:21,663][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:31:22,160][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:31:22,659][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:31:23,157][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:31:23,654][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:31:24,151][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:31:24,650][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:31:25,147][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:31:25,644][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:31:26,142][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:31:26,638][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:31:27,133][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:31:27,628][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:31:28,124][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:31:28,620][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21697 tokens. [2026-03-25 22:31:29,255][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.13%, Current % of VRAM taken: 60.49%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 22:31:29,995][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:31:29,997][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:31:29,999][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:31:30,836][__main__][INFO] - Iteration 86 took 1m 14s (9.56% Gen, 89.32% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 10m 49s. Estimated total time: 62h 7m 37s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 15s, 500 more iterations: 10h 21m 16s. [2026-03-25 22:31:30,839][__main__][INFO] - Starting iteration 86. [2026-03-25 22:31:31,238][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:31:31,239][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:31:38,372][__main__][INFO] - Number of regex retries in iteration 86: 0 [2026-03-25 22:31:38,373][__main__][INFO] - agents played in iteration 86 are Alice, Bob [2026-03-25 22:31:39,248][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:31:39,293][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:31:39,293][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:31:39,294][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:31:39,909][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:31:40,360][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:31:40,859][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:31:41,355][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:31:41,850][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:31:42,344][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:31:42,839][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:31:43,333][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:31:43,827][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:31:44,328][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:31:44,830][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:31:45,330][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:31:45,829][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:31:46,334][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:31:46,833][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:31:47,332][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:31:47,833][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:31:48,333][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:31:48,832][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:31:49,331][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:31:49,830][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:31:50,328][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:31:50,827][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:31:51,326][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:31:51,826][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:31:52,324][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:31:52,823][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:31:53,320][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:31:53,818][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:31:54,316][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:31:54,813][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:31:55,311][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:31:55,810][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:31:56,307][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:31:56,804][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:31:57,300][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:31:57,796][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:31:58,292][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:31:58,788][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:31:59,285][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:31:59,781][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:32:00,277][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:32:00,772][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:32:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:32:01,765][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:32:02,262][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:32:02,759][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:32:03,255][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:32:03,751][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:32:04,251][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:32:04,750][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:32:05,248][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:32:05,748][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:32:06,248][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:32:06,748][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:32:07,247][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:32:07,746][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:32:08,246][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:32:08,744][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:32:09,243][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:32:09,742][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:32:10,241][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:32:10,740][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:32:11,239][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:32:11,737][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:32:12,232][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:32:12,727][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:32:13,223][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:32:13,719][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:32:14,214][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:32:14,708][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:32:15,204][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:32:15,699][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:32:16,199][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:32:16,714][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:32:17,214][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:32:17,713][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:32:18,214][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:32:18,714][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:32:19,215][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:32:19,714][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:32:20,212][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:32:20,710][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:32:21,208][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:32:21,706][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:32:22,205][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:32:22,703][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:32:23,201][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:32:23,699][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:32:24,197][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:32:24,694][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:32:25,190][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:32:25,688][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:32:26,186][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:32:26,684][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:32:27,182][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:32:27,680][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:32:28,176][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:32:28,671][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:32:29,167][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:32:29,662][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:32:30,158][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:32:30,655][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:32:31,151][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:32:31,647][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:32:32,145][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:32:32,642][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:32:33,138][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:32:33,635][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:32:34,131][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:32:34,626][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:32:35,122][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:32:35,618][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:32:36,118][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:32:36,617][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:32:37,116][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:32:37,615][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:32:38,114][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:32:38,612][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:32:39,111][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:32:39,609][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:32:40,107][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:32:40,605][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:32:41,103][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:32:41,602][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:32:42,100][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:32:42,598][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:32:43,095][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:32:43,593][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21682 tokens. [2026-03-25 22:32:44,225][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-25 22:32:44,968][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:32:44,970][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:32:44,972][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:32:45,612][__main__][INFO] - Iteration 87 took 1m 14s (9.59% Gen, 89.55% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 0m 39s. Estimated total time: 61h 58m 42s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 57s, 500 more iterations: 10h 19m 47s. [2026-03-25 22:32:45,614][__main__][INFO] - Starting iteration 87. [2026-03-25 22:32:46,012][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:32:46,013][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:32:54,368][__main__][INFO] - Number of regex retries in iteration 87: 0 [2026-03-25 22:32:54,369][__main__][INFO] - agents played in iteration 87 are Alice, Bob [2026-03-25 22:32:55,251][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:32:55,295][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:32:55,296][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:32:55,296][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:32:55,901][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:32:56,357][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:32:56,861][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:32:57,362][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:32:57,861][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:32:58,360][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:32:58,859][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:32:59,358][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:32:59,857][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:33:00,355][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:33:00,853][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:33:01,352][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:33:01,849][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:33:02,347][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:33:02,844][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:33:03,341][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:33:03,839][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:33:04,337][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:33:04,834][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:33:05,331][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:33:05,829][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:33:06,327][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:33:06,826][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:33:07,325][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:33:07,824][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:33:08,324][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:33:08,824][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:33:09,323][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:33:09,822][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:33:10,320][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:33:10,818][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:33:11,316][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:33:11,814][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:33:12,312][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:33:12,811][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:33:13,309][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:33:13,808][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:33:14,306][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:33:14,805][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:33:15,304][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:33:15,804][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:33:16,302][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:33:16,802][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:33:17,301][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:33:17,799][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:33:18,298][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:33:18,797][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:33:19,295][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:33:19,792][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:33:20,289][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:33:20,788][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:33:21,288][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:33:21,786][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:33:22,286][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:33:22,784][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:33:23,283][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:33:23,781][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:33:24,278][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:33:24,777][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:33:25,275][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:33:25,772][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:33:26,270][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:33:26,768][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:33:27,266][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:33:27,764][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:33:28,263][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:33:28,761][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:33:29,261][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:33:29,760][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:33:30,259][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:33:30,757][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:33:31,254][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:33:31,752][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:33:32,249][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:33:32,747][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:33:33,246][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:33:33,744][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:33:34,242][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:33:34,741][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:33:35,239][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:33:35,737][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:33:36,234][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:33:36,731][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:33:37,228][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:33:37,726][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:33:38,224][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:33:38,721][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:33:39,219][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:33:39,717][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:33:40,215][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:33:40,713][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:33:41,210][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:33:41,707][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:33:42,206][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:33:42,705][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:33:43,203][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:33:43,701][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:33:44,199][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:33:44,697][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:33:45,195][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:33:45,693][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:33:46,190][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:33:46,688][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:33:47,186][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:33:47,684][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:33:48,184][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:33:48,681][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:33:49,178][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:33:49,676][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:33:50,173][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:33:50,671][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:33:51,170][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:33:51,668][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:33:52,168][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:33:52,667][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:33:53,166][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:33:53,664][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:33:54,163][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:33:54,662][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:33:55,162][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:33:55,662][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:33:56,161][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:33:56,659][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:33:57,157][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:33:57,656][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:33:58,154][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:33:58,653][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:33:59,151][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:33:59,650][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21687 tokens. [2026-03-25 22:34:00,278][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 22:34:01,006][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:34:01,008][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:34:01,010][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:34:01,652][__main__][INFO] - Iteration 88 took 1m 15s (11.05% Gen, 88.10% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 2m 41s. Estimated total time: 63h 2m 0s. Time estimates for 10 more iterations: 12m 36s, 100 more iterations: 2h 6m 4s, 500 more iterations: 10h 30m 20s. [2026-03-25 22:34:01,654][__main__][INFO] - Starting iteration 88. [2026-03-25 22:34:02,052][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:34:02,053][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:34:09,333][__main__][INFO] - Number of regex retries in iteration 88: 0 [2026-03-25 22:34:09,334][__main__][INFO] - agents played in iteration 88 are Alice, Bob [2026-03-25 22:34:10,212][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:34:10,256][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:34:10,257][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:34:10,258][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:34:10,862][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:34:11,319][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:34:11,823][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:34:12,324][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:34:12,824][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:34:13,323][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:34:13,823][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:34:14,323][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:34:14,823][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:34:15,329][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:34:15,834][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:34:16,336][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:34:16,841][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:34:17,342][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:34:17,845][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:34:18,348][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:34:18,850][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:34:19,349][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:34:19,847][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:34:20,345][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:34:20,844][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:34:21,343][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:34:21,842][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:34:22,341][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:34:22,839][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:34:23,336][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:34:23,834][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:34:24,332][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:34:24,831][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:34:25,330][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:34:25,829][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:34:26,327][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:34:26,826][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:34:27,326][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:34:27,826][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:34:28,328][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:34:28,842][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:34:29,346][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:34:29,846][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:34:30,346][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:34:30,846][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:34:31,343][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:34:31,842][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:34:32,340][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:34:32,838][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:34:33,334][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:34:33,833][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:34:34,332][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:34:34,829][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:34:35,327][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:34:35,825][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:34:36,322][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:34:36,821][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:34:37,320][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:34:37,820][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:34:38,317][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:34:38,815][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:34:39,311][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:34:39,809][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:34:40,307][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:34:40,804][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:34:41,302][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:34:41,800][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:34:42,297][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:34:42,794][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:34:43,293][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:34:43,793][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:34:44,290][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:34:44,789][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:34:45,288][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:34:45,787][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:34:46,285][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:34:46,784][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:34:47,285][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:34:47,787][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:34:48,291][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:34:48,792][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:34:49,294][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:34:49,795][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:34:50,296][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:34:50,798][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:34:51,295][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:34:51,793][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:34:52,290][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:34:52,790][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:34:53,288][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:34:53,785][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:34:54,284][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:34:54,781][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:34:55,281][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:34:55,780][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:34:56,277][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:34:56,775][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:34:57,272][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:34:57,770][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:34:58,267][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:34:58,765][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:34:59,264][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:34:59,765][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:35:00,266][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:35:00,766][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:35:01,266][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:35:01,765][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:35:02,264][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:35:02,764][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:35:03,263][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:35:03,763][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:35:04,262][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:35:04,762][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:35:05,260][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:35:05,758][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:35:06,256][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:35:06,754][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:35:07,251][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:35:07,749][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:35:08,248][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:35:08,746][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:35:09,244][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:35:09,742][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:35:10,240][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:35:10,737][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:35:11,234][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:35:11,732][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:35:12,229][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:35:12,727][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:35:13,224][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:35:13,721][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:35:14,219][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:35:14,716][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21563 tokens. [2026-03-25 22:35:15,347][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-25 22:35:16,109][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:35:16,112][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:35:16,113][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:35:16,920][__main__][INFO] - Iteration 89 took 1m 14s (9.72% Gen, 89.20% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 22m 49s. Estimated total time: 62h 23m 24s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 46s, 500 more iterations: 10h 23m 54s. [2026-03-25 22:35:16,922][__main__][INFO] - Starting iteration 89. [2026-03-25 22:35:17,324][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:35:17,325][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:35:20,914][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:35:25,513][__main__][INFO] - Number of regex retries in iteration 89: 1 [2026-03-25 22:35:25,514][__main__][INFO] - agents played in iteration 89 are Alice, Bob [2026-03-25 22:35:26,395][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:35:26,721][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:35:26,722][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:35:26,723][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:35:27,333][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:35:27,790][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:35:28,293][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:35:28,792][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:35:29,291][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:35:29,789][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:35:30,289][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:35:30,789][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:35:31,289][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:35:31,788][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:35:32,288][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:35:32,786][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:35:33,284][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:35:33,783][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:35:34,283][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:35:34,781][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:35:35,280][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:35:35,778][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:35:36,276][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:35:36,774][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:35:37,272][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:35:37,771][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:35:38,270][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:35:38,769][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:35:39,267][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:35:39,766][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:35:40,264][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:35:40,763][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:35:41,261][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:35:41,759][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:35:42,258][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:35:42,756][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:35:43,254][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:35:43,752][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:35:44,249][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:35:44,747][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:35:45,244][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:35:45,745][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:35:46,243][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:35:46,743][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:35:47,241][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:35:47,740][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:35:48,241][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:35:48,740][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:35:49,238][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:35:49,736][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:35:50,235][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:35:50,733][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:35:51,231][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:35:51,730][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:35:52,230][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:35:52,728][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:35:53,227][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:35:53,726][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:35:54,225][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:35:54,725][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:35:55,224][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:35:55,725][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:35:56,225][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:35:56,724][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:35:57,224][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:35:57,724][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:35:58,225][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:35:58,725][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:35:59,225][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:35:59,724][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:36:00,224][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:36:00,724][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:36:01,224][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:36:01,724][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:36:02,224][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:36:02,723][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:36:03,222][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:36:03,721][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:36:04,221][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:36:04,720][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:36:05,220][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:36:05,719][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:36:06,219][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:36:06,718][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:36:07,217][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:36:07,716][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:36:08,214][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:36:08,712][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:36:09,210][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:36:09,711][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:36:10,210][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:36:10,709][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:36:11,208][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:36:11,706][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:36:12,205][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:36:12,704][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:36:13,204][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:36:13,702][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:36:14,201][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:36:14,699][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:36:15,197][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:36:15,695][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:36:16,192][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:36:16,689][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:36:17,187][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:36:17,686][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:36:18,185][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:36:18,684][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:36:19,182][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:36:19,680][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:36:20,179][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:36:20,676][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:36:21,175][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:36:21,673][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:36:22,171][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:36:22,670][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:36:23,169][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:36:23,667][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:36:24,165][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:36:24,663][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:36:25,163][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:36:25,662][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:36:26,161][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:36:26,660][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:36:27,159][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:36:27,658][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:36:28,156][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:36:28,654][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:36:29,153][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:36:29,652][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:36:30,151][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:36:30,651][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:36:31,150][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21635 tokens. [2026-03-25 22:36:31,784][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-25 22:36:32,516][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:36:32,519][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:36:32,520][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:36:33,164][__main__][INFO] - Iteration 90 took 1m 15s (10.80% Gen, 88.35% Train). Generation: 8s, Training: 1m 7s. Estimated remaining time: 61h 10m 9s. Estimated total time: 63h 12m 0s. Time estimates for 10 more iterations: 12m 38s, 100 more iterations: 2h 6m 24s, 500 more iterations: 10h 32m 0s. [2026-03-25 22:36:33,166][__main__][INFO] - Starting iteration 90. [2026-03-25 22:36:33,564][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:36:33,565][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:36:41,002][__main__][INFO] - Number of regex retries in iteration 90: 0 [2026-03-25 22:36:41,003][__main__][INFO] - agents played in iteration 90 are Alice, Bob [2026-03-25 22:36:41,886][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:36:41,930][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:36:41,931][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:36:41,931][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:36:42,542][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:36:42,997][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:36:43,501][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:36:44,000][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:36:44,498][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:36:44,996][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:36:45,494][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:36:45,993][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:36:46,491][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:36:46,991][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:36:47,490][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:36:47,991][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:36:48,492][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:36:48,992][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:36:49,491][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:36:49,990][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:36:50,489][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:36:50,987][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:36:51,484][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:36:51,983][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:36:52,482][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:36:52,980][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:36:53,478][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:36:53,976][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:36:54,473][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:36:54,971][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:36:55,470][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:36:55,969][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:36:56,466][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:36:56,965][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:36:57,463][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:36:57,962][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:36:58,459][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:36:58,958][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:36:59,456][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:36:59,954][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:37:00,452][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:37:00,951][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:37:01,449][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:37:01,947][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:37:02,446][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:37:02,944][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:37:03,443][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:37:03,941][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:37:04,442][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:37:04,940][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:37:05,439][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:37:05,939][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:37:06,437][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:37:06,935][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:37:07,434][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:37:07,932][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:37:08,431][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:37:08,931][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:37:09,431][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:37:09,930][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:37:10,429][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:37:10,928][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:37:11,427][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:37:11,925][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:37:12,424][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:37:12,923][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:37:13,422][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:37:13,921][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:37:14,419][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:37:14,917][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:37:15,417][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:37:15,917][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:37:16,416][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:37:16,914][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:37:17,414][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:37:17,911][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:37:18,411][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:37:18,910][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:37:19,410][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:37:19,909][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:37:20,408][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:37:20,907][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:37:21,407][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:37:21,907][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:37:22,406][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:37:22,906][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:37:23,405][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:37:23,906][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:37:24,405][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:37:24,904][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:37:25,403][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:37:25,902][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:37:26,400][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:37:26,897][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:37:27,393][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:37:27,890][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:37:28,387][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:37:28,884][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:37:29,382][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:37:29,880][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:37:30,377][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:37:30,875][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:37:31,373][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:37:31,868][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:37:32,367][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:37:32,864][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:37:33,363][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:37:33,862][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:37:34,361][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:37:34,860][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:37:35,358][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:37:35,856][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:37:36,355][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:37:36,854][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:37:37,352][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:37:37,851][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:37:38,350][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:37:38,850][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:37:39,348][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:37:39,847][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:37:40,346][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:37:40,845][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:37:41,345][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:37:41,844][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:37:42,343][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:37:42,842][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:37:43,340][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:37:43,837][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:37:44,336][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:37:44,835][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:37:45,333][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:37:45,831][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:37:46,330][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21616 tokens. [2026-03-25 22:37:46,964][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:37:47,691][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:37:47,693][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:37:47,695][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:37:48,341][__main__][INFO] - Iteration 91 took 1m 14s (9.95% Gen, 89.19% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 15m 46s. Estimated total time: 62h 18m 52s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 37s, 500 more iterations: 10h 23m 8s. [2026-03-25 22:37:48,344][__main__][INFO] - Starting iteration 91. [2026-03-25 22:37:48,743][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:37:48,744][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:37:55,350][__main__][INFO] - Number of regex retries in iteration 91: 0 [2026-03-25 22:37:55,351][__main__][INFO] - agents played in iteration 91 are Alice, Bob [2026-03-25 22:37:56,239][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:37:56,562][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:37:56,562][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:37:56,563][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:37:57,170][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:37:57,625][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:37:58,127][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:37:58,626][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:37:59,124][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:37:59,623][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:38:00,122][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:38:00,620][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:38:01,118][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:38:01,616][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:38:02,114][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:38:02,612][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:38:03,111][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:38:03,610][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:38:04,108][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:38:04,607][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:38:05,105][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:38:05,606][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:38:06,104][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:38:06,605][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:38:07,104][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:38:07,605][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:38:08,105][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:38:08,604][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:38:09,104][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:38:09,602][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:38:10,103][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:38:10,602][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:38:11,102][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:38:11,601][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:38:12,099][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:38:12,597][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:38:13,094][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:38:13,592][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:38:14,090][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:38:14,589][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:38:15,088][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:38:15,586][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:38:16,084][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:38:16,583][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:38:17,081][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:38:17,578][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:38:18,076][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:38:18,576][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:38:19,075][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:38:19,573][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:38:20,071][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:38:20,569][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:38:21,067][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:38:21,567][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:38:22,066][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:38:22,565][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:38:23,065][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:38:23,564][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:38:24,063][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:38:24,562][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:38:25,061][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:38:25,557][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:38:26,053][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:38:26,550][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:38:27,047][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:38:27,544][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:38:28,041][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:38:28,538][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:38:29,035][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:38:29,533][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:38:30,030][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:38:30,529][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:38:31,027][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:38:31,525][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:38:32,023][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:38:32,522][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:38:33,022][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:38:33,519][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:38:34,016][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:38:34,513][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:38:35,011][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:38:35,509][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:38:36,007][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:38:36,506][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:38:37,004][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:38:37,503][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:38:38,001][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:38:38,500][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:38:38,998][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:38:39,496][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:38:39,994][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:38:40,493][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:38:40,992][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:38:41,489][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:38:41,988][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:38:42,486][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:38:42,985][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:38:43,483][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:38:43,984][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:38:44,483][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:38:44,982][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:38:45,483][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:38:45,984][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:38:46,482][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:38:46,980][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:38:47,478][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:38:47,977][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:38:48,478][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:38:48,976][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:38:49,474][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:38:49,972][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:38:50,470][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:38:50,968][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:38:51,466][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:38:51,965][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:38:52,463][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:38:52,963][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:38:53,463][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:38:53,961][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:38:54,460][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:38:54,958][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:38:55,457][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:38:55,956][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:38:56,454][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:38:56,954][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:38:57,451][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:38:57,948][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:38:58,444][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:38:58,941][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:38:59,437][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:38:59,933][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:39:00,429][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:39:00,927][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-25 22:39:01,560][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.13%, Current % of VRAM taken: 60.49%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 22:39:02,291][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:39:02,294][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:39:02,295][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:39:02,939][__main__][INFO] - Iteration 92 took 1m 14s (8.91% Gen, 90.23% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 45m 28s. Estimated total time: 61h 49m 48s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 18s. [2026-03-25 22:39:02,942][__main__][INFO] - Starting iteration 92. [2026-03-25 22:39:03,345][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:39:03,345][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:39:07,857][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:39:10,266][__main__][INFO] - Number of regex retries in iteration 92: 1 [2026-03-25 22:39:10,267][__main__][INFO] - agents played in iteration 92 are Alice, Bob [2026-03-25 22:39:11,148][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:39:11,193][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:39:11,193][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:39:11,194][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:39:11,797][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:39:12,254][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:39:12,758][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:39:13,259][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:39:13,759][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:39:14,260][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:39:14,760][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:39:15,260][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:39:15,760][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:39:16,262][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:39:16,762][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:39:17,262][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:39:17,762][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:39:18,262][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:39:18,763][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:39:19,265][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:39:19,765][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:39:20,263][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:39:20,761][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:39:21,259][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:39:21,757][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:39:22,255][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:39:22,754][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:39:23,252][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:39:23,750][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:39:24,248][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:39:24,745][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:39:25,243][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:39:25,741][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:39:26,238][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:39:26,735][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:39:27,232][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:39:27,731][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:39:28,229][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:39:28,728][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:39:29,226][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:39:29,724][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:39:30,221][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:39:30,719][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:39:31,216][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:39:31,713][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:39:32,211][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:39:32,710][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:39:33,210][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:39:33,709][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:39:34,208][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:39:34,706][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:39:35,206][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:39:35,705][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:39:36,204][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:39:36,703][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:39:37,202][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:39:37,701][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:39:38,196][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:39:38,694][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:39:39,192][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:39:39,689][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:39:40,188][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:39:40,687][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:39:41,185][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:39:41,684][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:39:42,183][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:39:42,682][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:39:43,181][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:39:43,678][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:39:44,177][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:39:44,676][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:39:45,176][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:39:45,675][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:39:46,174][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:39:46,674][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:39:47,174][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:39:47,674][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:39:48,175][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:39:48,676][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:39:49,176][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:39:49,676][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:39:50,176][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:39:50,675][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:39:51,175][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:39:51,674][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:39:52,171][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:39:52,669][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:39:53,166][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:39:53,663][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:39:54,160][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:39:54,657][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:39:55,154][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:39:55,651][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:39:56,149][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:39:56,647][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:39:57,145][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:39:57,643][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:39:58,141][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:39:58,639][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:39:59,136][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:39:59,634][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:40:00,133][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:40:00,632][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:40:01,130][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:40:01,629][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:40:02,127][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:40:02,624][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:40:03,122][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:40:03,620][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:40:04,119][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:40:04,617][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:40:05,115][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:40:05,613][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:40:06,113][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:40:06,611][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:40:07,110][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:40:07,609][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:40:08,107][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:40:08,606][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:40:09,104][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:40:09,603][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:40:10,101][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:40:10,598][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:40:11,095][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:40:11,594][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:40:12,092][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:40:12,590][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:40:13,089][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:40:13,588][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:40:14,087][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:40:14,585][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:40:15,083][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:40:15,585][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21713 tokens. [2026-03-25 22:40:16,210][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-25 22:40:16,941][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:40:16,944][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:40:16,945][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:40:17,649][__main__][INFO] - Iteration 93 took 1m 14s (9.31% Gen, 89.74% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 49m 38s. Estimated total time: 61h 55m 13s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 50s, 500 more iterations: 10h 19m 12s. [2026-03-25 22:40:17,651][__main__][INFO] - Starting iteration 93. [2026-03-25 22:40:18,054][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:40:18,055][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:40:18,634][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:40:25,564][__main__][INFO] - Number of regex retries in iteration 93: 1 [2026-03-25 22:40:25,565][__main__][INFO] - agents played in iteration 93 are Alice, Bob [2026-03-25 22:40:26,441][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:40:26,486][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:40:26,486][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:40:26,487][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:40:27,095][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:40:27,550][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:40:28,052][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:40:28,551][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:40:29,049][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:40:29,550][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:40:30,050][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:40:30,549][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:40:31,049][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:40:31,547][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:40:32,045][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:40:32,543][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:40:33,043][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:40:33,542][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:40:34,041][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:40:34,540][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:40:35,038][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:40:35,536][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:40:36,034][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:40:36,533][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:40:37,032][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:40:37,531][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:40:38,029][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:40:38,528][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:40:39,026][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:40:39,527][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:40:40,027][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:40:40,526][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:40:41,025][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:40:41,524][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:40:42,024][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:40:42,523][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:40:43,023][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:40:43,521][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:40:44,018][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:40:44,517][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:40:45,015][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:40:45,512][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:40:46,011][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:40:46,510][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:40:47,008][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:40:47,507][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:40:48,006][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:40:48,506][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:40:49,005][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:40:49,504][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:40:50,002][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:40:50,500][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:40:50,998][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:40:51,496][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:40:51,994][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:40:52,492][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:40:52,991][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:40:53,490][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:40:53,988][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:40:54,487][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:40:54,985][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:40:55,486][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:40:55,985][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:40:56,483][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:40:56,982][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:40:57,482][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:40:57,982][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:40:58,481][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:40:58,980][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:40:59,479][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:40:59,978][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:41:00,473][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:41:00,973][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:41:01,473][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:41:01,971][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:41:02,469][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:41:02,967][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:41:03,465][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:41:03,963][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:41:04,461][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:41:04,960][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:41:05,458][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:41:05,956][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:41:06,454][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:41:06,952][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:41:07,451][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:41:07,950][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:41:08,448][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:41:08,947][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:41:09,446][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:41:09,945][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:41:10,444][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:41:10,942][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:41:11,442][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:41:11,940][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:41:12,438][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:41:12,936][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:41:13,435][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:41:13,933][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:41:14,433][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:41:14,932][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:41:15,430][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:41:15,930][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:41:16,428][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:41:16,927][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:41:17,425][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:41:17,925][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:41:18,425][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:41:18,923][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:41:19,422][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:41:19,920][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:41:20,417][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:41:20,915][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:41:21,413][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:41:21,911][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:41:22,410][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:41:22,908][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:41:23,407][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:41:23,905][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:41:24,405][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:41:24,904][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:41:25,406][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:41:25,905][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:41:26,404][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:41:26,903][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:41:27,402][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:41:27,901][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:41:28,400][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:41:28,900][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:41:29,399][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:41:29,898][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:41:30,398][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:41:30,896][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21729 tokens. [2026-03-25 22:41:31,543][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 22:41:32,272][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:41:32,274][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:41:32,276][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:41:32,917][__main__][INFO] - Iteration 94 took 1m 14s (10.03% Gen, 89.11% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 16m 20s. Estimated total time: 62h 23m 11s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 46s, 500 more iterations: 10h 23m 51s. [2026-03-25 22:41:32,920][__main__][INFO] - Starting iteration 94. [2026-03-25 22:41:33,320][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:41:33,321][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:41:39,031][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:41:41,757][__main__][INFO] - Number of regex retries in iteration 94: 1 [2026-03-25 22:41:41,758][__main__][INFO] - agents played in iteration 94 are Alice, Bob [2026-03-25 22:41:42,642][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:41:42,687][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:41:42,688][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:41:42,688][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:41:43,305][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:41:43,760][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:41:44,262][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:41:44,760][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:41:45,259][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:41:45,760][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:41:46,260][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:41:46,758][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:41:47,257][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:41:47,758][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:41:48,260][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:41:48,759][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:41:49,258][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:41:49,757][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:41:50,256][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:41:50,755][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:41:51,256][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:41:51,754][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:41:52,253][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:41:52,751][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:41:53,250][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:41:53,748][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:41:54,247][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:41:54,746][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:41:55,246][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:41:55,745][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:41:56,244][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:41:56,744][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:41:57,243][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:41:57,744][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:41:58,244][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:41:58,744][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:41:59,244][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:41:59,743][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:42:00,241][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:42:00,740][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:42:01,240][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:42:01,738][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:42:02,237][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:42:02,737][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:42:03,235][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:42:03,734][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:42:04,233][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:42:04,732][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:42:05,232][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:42:05,729][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:42:06,228][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:42:06,728][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:42:07,227][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:42:07,729][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:42:08,230][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:42:08,731][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:42:09,232][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:42:09,732][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:42:10,232][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:42:10,733][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:42:11,232][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:42:11,732][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:42:12,231][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:42:12,730][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:42:13,228][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:42:13,727][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:42:14,226][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:42:14,725][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:42:15,223][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:42:15,721][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:42:16,219][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:42:16,717][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:42:17,215][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:42:17,713][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:42:18,211][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:42:18,711][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:42:19,210][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:42:19,709][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:42:20,208][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:42:20,708][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:42:21,207][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:42:21,706][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:42:22,205][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:42:22,704][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:42:23,201][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:42:23,695][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:42:24,189][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:42:24,687][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:42:25,179][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:42:25,677][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:42:26,171][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:42:26,671][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:42:27,165][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:42:27,664][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:42:28,162][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:42:28,661][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:42:29,159][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:42:29,658][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:42:30,156][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:42:30,656][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:42:31,154][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:42:31,653][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:42:32,151][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:42:32,650][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:42:33,150][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:42:33,648][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:42:34,148][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:42:34,647][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:42:35,147][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:42:35,646][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:42:36,144][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:42:36,644][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:42:37,142][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:42:37,641][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:42:38,141][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:42:38,639][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:42:39,139][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:42:39,638][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:42:40,139][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:42:40,638][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:42:41,138][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:42:41,638][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:42:42,138][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:42:42,638][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:42:43,138][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:42:43,637][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:42:44,136][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:42:44,634][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:42:45,134][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:42:45,634][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:42:46,134][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:42:46,633][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:42:47,131][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21447 tokens. [2026-03-25 22:42:47,794][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-25 22:42:48,541][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:42:48,544][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:42:48,545][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:42:49,244][__main__][INFO] - Iteration 95 took 1m 15s (11.11% Gen, 87.97% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 61h 8m 7s. Estimated total time: 63h 16m 14s. Time estimates for 10 more iterations: 12m 39s, 100 more iterations: 2h 6m 32s, 500 more iterations: 10h 32m 42s. [2026-03-25 22:42:49,246][__main__][INFO] - Starting iteration 95. [2026-03-25 22:42:49,652][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:42:49,653][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:42:56,769][__main__][INFO] - Number of regex retries in iteration 95: 0 [2026-03-25 22:42:56,770][__main__][INFO] - agents played in iteration 95 are Alice, Bob [2026-03-25 22:42:57,646][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:42:57,691][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:42:57,692][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:42:57,692][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:42:58,316][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:42:58,774][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:42:59,278][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:42:59,779][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:43:00,279][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:43:00,779][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:43:01,279][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:43:01,779][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:43:02,279][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:43:02,779][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:43:03,278][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:43:03,778][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:43:04,278][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:43:04,776][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:43:05,275][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:43:05,774][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:43:06,274][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:43:06,772][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:43:07,270][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:43:07,769][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:43:08,268][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:43:08,765][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:43:09,263][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:43:09,760][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:43:10,258][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:43:10,757][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:43:11,256][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:43:11,755][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:43:12,254][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:43:12,753][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:43:13,252][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:43:13,751][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:43:14,249][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:43:14,748][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:43:15,246][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:43:15,748][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:43:16,248][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:43:16,746][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:43:17,244][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:43:17,742][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:43:18,243][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:43:18,745][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:43:19,245][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:43:19,744][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:43:20,243][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:43:20,743][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:43:21,241][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:43:21,740][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:43:22,240][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:43:22,739][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:43:23,238][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:43:23,736][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:43:24,235][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:43:24,732][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:43:25,231][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:43:25,731][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:43:26,230][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:43:26,727][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:43:27,224][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:43:27,720][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:43:28,217][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:43:28,714][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:43:29,210][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:43:29,708][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:43:30,206][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:43:30,707][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:43:31,208][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:43:31,707][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:43:32,207][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:43:32,706][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:43:33,207][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:43:33,707][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:43:34,207][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:43:34,707][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:43:35,206][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:43:35,704][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:43:36,202][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:43:36,700][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:43:37,197][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:43:37,695][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:43:38,194][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:43:38,692][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:43:39,190][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:43:39,687][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:43:40,184][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:43:40,681][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:43:41,180][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:43:41,676][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:43:42,174][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:43:42,673][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:43:43,173][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:43:43,673][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:43:44,173][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:43:44,672][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:43:45,171][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:43:45,670][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:43:46,170][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:43:46,668][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:43:47,166][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:43:47,665][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:43:48,166][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:43:48,665][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:43:49,162][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:43:49,659][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:43:50,157][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:43:50,656][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:43:51,155][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:43:51,654][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:43:52,154][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:43:52,653][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:43:53,152][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:43:53,652][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:43:54,151][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:43:54,651][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:43:55,150][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:43:55,649][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:43:56,147][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:43:56,645][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:43:57,143][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:43:57,642][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:43:58,141][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:43:58,637][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:43:59,134][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:43:59,631][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:44:00,128][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:44:00,626][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:44:01,122][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:44:01,620][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:44:02,116][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21747 tokens. [2026-03-25 22:44:02,770][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 22:44:03,513][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:44:03,516][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:44:03,517][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:44:04,460][__main__][INFO] - Iteration 96 took 1m 14s (9.51% Gen, 89.22% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 11m 1s. Estimated total time: 62h 20m 23s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 40s, 500 more iterations: 10h 23m 23s. [2026-03-25 22:44:04,462][__main__][INFO] - Starting iteration 96. [2026-03-25 22:44:05,221][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:44:05,221][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:44:06,957][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:44:13,047][__main__][INFO] - Number of regex retries in iteration 96: 1 [2026-03-25 22:44:13,048][__main__][INFO] - agents played in iteration 96 are Alice, Bob [2026-03-25 22:44:13,929][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:44:13,973][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:44:13,974][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:44:13,974][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:44:14,581][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:44:15,037][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:44:15,540][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:44:16,039][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:44:16,536][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:44:17,034][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:44:17,533][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:44:18,032][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:44:18,533][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:44:19,034][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:44:19,534][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:44:20,032][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:44:20,532][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:44:21,031][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:44:21,531][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:44:22,031][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:44:22,530][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:44:23,029][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:44:23,529][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:44:24,029][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:44:24,528][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:44:25,028][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:44:25,528][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:44:26,028][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:44:26,526][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:44:27,027][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:44:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:44:28,028][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:44:28,529][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:44:29,029][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:44:29,529][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:44:30,029][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:44:30,529][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:44:31,029][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:44:31,528][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:44:32,027][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:44:32,526][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:44:33,026][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:44:33,525][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:44:34,023][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:44:34,521][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:44:35,020][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:44:35,519][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:44:36,018][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:44:36,516][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:44:37,015][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:44:37,514][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:44:38,011][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:44:38,510][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:44:39,008][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:44:39,507][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:44:40,007][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:44:40,506][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:44:41,005][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:44:41,503][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:44:42,002][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:44:42,500][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:44:42,999][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:44:43,498][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:44:43,998][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:44:44,497][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:44:44,998][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:44:45,500][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:44:46,001][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:44:46,500][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:44:46,998][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:44:47,496][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:44:47,995][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:44:48,496][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:44:48,994][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:44:49,491][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:44:49,990][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:44:50,487][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:44:50,986][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:44:51,486][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:44:51,986][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:44:52,486][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:44:52,985][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:44:53,484][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:44:53,982][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:44:54,481][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:44:54,981][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:44:55,480][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:44:55,978][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:44:56,476][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:44:56,975][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:44:57,473][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:44:57,971][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:44:58,470][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:44:58,969][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:44:59,469][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:44:59,968][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:45:00,467][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:45:00,965][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:45:01,464][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:45:01,962][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:45:02,460][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:45:02,958][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:45:03,456][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:45:03,954][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:45:04,453][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:45:04,952][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:45:05,450][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:45:05,950][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:45:06,449][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:45:06,949][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:45:07,448][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:45:07,947][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:45:08,447][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:45:08,946][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:45:09,446][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:45:09,944][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:45:10,443][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:45:10,941][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:45:11,439][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:45:11,937][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:45:12,435][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:45:12,932][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:45:13,430][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:45:13,928][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:45:14,426][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:45:14,927][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:45:15,427][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:45:15,927][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:45:16,427][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:45:16,927][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:45:17,427][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:45:17,928][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:45:18,429][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21692 tokens. [2026-03-25 22:45:19,065][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:45:19,800][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:45:19,803][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:45:19,804][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:45:20,647][__main__][INFO] - Iteration 97 took 1m 15s (10.38% Gen, 88.50% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 40m 42s. Estimated total time: 62h 51m 20s. Time estimates for 10 more iterations: 12m 34s, 100 more iterations: 2h 5m 42s, 500 more iterations: 10h 28m 33s. [2026-03-25 22:45:20,650][__main__][INFO] - Starting iteration 97. [2026-03-25 22:45:21,054][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:45:21,055][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:45:26,674][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:45:27,731][__main__][INFO] - Number of regex retries in iteration 97: 1 [2026-03-25 22:45:27,732][__main__][INFO] - agents played in iteration 97 are Alice, Bob [2026-03-25 22:45:28,611][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:45:28,655][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:45:28,656][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:45:28,656][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:45:29,264][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:45:29,721][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:45:30,226][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:45:30,727][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:45:31,228][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:45:31,727][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:45:32,228][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:45:32,728][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:45:33,228][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:45:33,729][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:45:34,230][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:45:34,731][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:45:35,230][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:45:35,731][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:45:36,231][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:45:36,732][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:45:37,232][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:45:37,731][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:45:38,229][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:45:38,727][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:45:39,227][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:45:39,726][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:45:40,283][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:45:40,789][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:45:41,286][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:45:41,784][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:45:42,283][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:45:42,780][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:45:43,278][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:45:43,776][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:45:44,273][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:45:44,772][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:45:45,270][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:45:45,768][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:45:46,267][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:45:46,766][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:45:47,264][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:45:47,762][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:45:48,263][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:45:48,763][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:45:49,261][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:45:49,760][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:45:50,258][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:45:50,756][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:45:51,255][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:45:51,754][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:45:52,253][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:45:52,751][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:45:53,251][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:45:53,749][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:45:54,246][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:45:54,745][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:45:55,243][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:45:55,740][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:45:56,238][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:45:56,737][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:45:57,235][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:45:57,737][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:45:58,239][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:45:58,741][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:45:59,244][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:45:59,746][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:46:00,248][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:46:00,750][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:46:01,253][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:46:01,752][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:46:02,252][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:46:02,751][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:46:03,250][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:46:03,750][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:46:04,249][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:46:04,749][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:46:05,249][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:46:05,749][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:46:06,250][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:46:06,752][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:46:07,253][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:46:07,753][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:46:08,253][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:46:08,753][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:46:09,253][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:46:09,752][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:46:10,250][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:46:10,748][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:46:11,246][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:46:11,745][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:46:12,243][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:46:12,740][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:46:13,238][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:46:13,736][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:46:14,235][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:46:14,735][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:46:15,234][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:46:15,736][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:46:16,235][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:46:16,734][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:46:17,233][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:46:17,732][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:46:18,234][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:46:18,734][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:46:19,234][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:46:19,732][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:46:20,231][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:46:20,729][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:46:21,230][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:46:21,728][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:46:22,226][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:46:22,723][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:46:23,220][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:46:23,717][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:46:24,214][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:46:24,711][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:46:25,210][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:46:25,709][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:46:26,207][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:46:26,706][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:46:27,205][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:46:27,703][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:46:28,201][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:46:28,699][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:46:29,197][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:46:29,698][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:46:30,199][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:46:30,700][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:46:31,202][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:46:31,704][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:46:32,208][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:46:32,710][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:46:33,210][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21742 tokens. [2026-03-25 22:46:33,859][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.67%, Block Peak % of device VRAM: 62.53%, ΔTime: 00:01:04 [2026-03-25 22:46:34,494][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:46:34,497][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:46:34,498][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:46:35,144][__main__][INFO] - Iteration 98 took 1m 14s (9.01% Gen, 90.12% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 32m 38s. Estimated total time: 61h 44m 31s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 29s, 500 more iterations: 10h 17m 25s. [2026-03-25 22:46:35,148][__main__][INFO] - Starting iteration 98. [2026-03-25 22:46:35,550][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:46:35,551][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:46:42,592][__main__][INFO] - Number of regex retries in iteration 98: 0 [2026-03-25 22:46:42,593][__main__][INFO] - agents played in iteration 98 are Alice, Bob [2026-03-25 22:46:43,475][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:46:43,801][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:46:43,801][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:46:43,802][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:46:44,410][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:46:44,867][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:46:45,371][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:46:45,870][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:46:46,370][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:46:46,871][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:46:47,370][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:46:47,870][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:46:48,372][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:46:48,871][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:46:49,368][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:46:49,866][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:46:50,364][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:46:50,861][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:46:51,358][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:46:51,856][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:46:52,354][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:46:52,854][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:46:53,353][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:46:53,853][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:46:54,352][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:46:54,852][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:46:55,352][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:46:55,851][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:46:56,350][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:46:56,848][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:46:57,347][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:46:57,846][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:46:58,345][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:46:58,844][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:46:59,341][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:46:59,840][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:47:00,337][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:47:00,836][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:47:01,334][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:47:01,833][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:47:02,332][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:47:02,832][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:47:03,330][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:47:03,829][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:47:04,329][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:47:04,830][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:47:05,331][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:47:05,832][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:47:06,333][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:47:06,834][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:47:07,334][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:47:07,834][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:47:08,336][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:47:08,837][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:47:09,336][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:47:09,835][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:47:10,334][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:47:10,834][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:47:11,334][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:47:11,833][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:47:12,333][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:47:12,833][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:47:13,332][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:47:13,831][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:47:14,332][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:47:14,832][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:47:15,331][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:47:15,831][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:47:16,330][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:47:16,830][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:47:17,330][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:47:17,829][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:47:18,331][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:47:18,833][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:47:19,333][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:47:19,834][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:47:20,339][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:47:20,836][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:47:21,333][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:47:21,831][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:47:22,329][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:47:22,828][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:47:23,326][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:47:23,825][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:47:24,323][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:47:24,821][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:47:25,321][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:47:25,820][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:47:26,318][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:47:26,817][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:47:27,315][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:47:27,816][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:47:28,315][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:47:28,814][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:47:29,313][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:47:29,812][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:47:30,310][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:47:30,809][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:47:31,308][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:47:31,806][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:47:32,304][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:47:32,803][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:47:33,302][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:47:33,799][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:47:34,297][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:47:34,796][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:47:35,296][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:47:35,794][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:47:36,293][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:47:36,793][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:47:37,293][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:47:37,792][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:47:38,292][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:47:38,794][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:47:39,294][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:47:39,793][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:47:40,293][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:47:40,793][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:47:41,292][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:47:41,791][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:47:42,290][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:47:42,790][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:47:43,289][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:47:43,789][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:47:44,288][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:47:44,788][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:47:45,289][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:47:45,792][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:47:46,293][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:47:46,793][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:47:47,293][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:47:47,793][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:47:48,295][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21730 tokens. [2026-03-25 22:47:48,937][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 22:47:49,678][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:47:49,680][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:47:49,682][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:47:50,339][__main__][INFO] - Iteration 99 took 1m 14s (9.42% Gen, 89.70% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 60h 6m 20s. Estimated total time: 62h 19m 28s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 38s, 500 more iterations: 10h 23m 14s. [2026-03-25 22:47:50,342][__main__][INFO] - Starting iteration 99. [2026-03-25 22:47:50,751][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:47:50,751][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:47:58,331][__main__][INFO] - Number of regex retries in iteration 99: 0 [2026-03-25 22:47:58,332][__main__][INFO] - agents played in iteration 99 are Alice, Bob [2026-03-25 22:47:59,210][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:47:59,255][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:47:59,255][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:47:59,256][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:47:59,864][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:48:00,321][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:48:00,826][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:48:01,327][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:48:01,827][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:48:02,326][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:48:02,825][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:48:03,324][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:48:03,822][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:48:04,321][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:48:04,817][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:48:05,318][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:48:05,816][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:48:06,316][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:48:06,815][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:48:07,313][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:48:07,813][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:48:08,312][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:48:08,811][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:48:09,310][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:48:09,808][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:48:10,306][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:48:10,805][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:48:11,303][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:48:11,801][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:48:12,299][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:48:12,797][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:48:13,295][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:48:13,794][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:48:14,293][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:48:14,791][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:48:15,289][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:48:15,787][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:48:16,287][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:48:16,787][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:48:17,287][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:48:17,787][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:48:18,289][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:48:18,788][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:48:19,288][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:48:19,788][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:48:20,286][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:48:20,785][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:48:21,284][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:48:21,781][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:48:22,279][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:48:22,776][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:48:23,274][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:48:23,771][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:48:24,270][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:48:24,770][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:48:25,270][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:48:25,768][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:48:26,267][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:48:26,767][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:48:27,266][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:48:27,764][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:48:28,266][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:48:28,768][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:48:29,269][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:48:29,770][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:48:30,270][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:48:30,770][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:48:31,271][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:48:31,771][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:48:32,269][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:48:32,769][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:48:33,267][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:48:33,765][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:48:34,264][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:48:34,761][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:48:35,260][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:48:35,758][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:48:36,256][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:48:36,755][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:48:37,253][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:48:37,753][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:48:38,252][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:48:38,751][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:48:39,249][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:48:39,748][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:48:40,247][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:48:40,746][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:48:41,244][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:48:41,743][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:48:42,241][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:48:42,739][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:48:43,236][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:48:43,734][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:48:44,233][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:48:44,733][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:48:45,232][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:48:45,730][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:48:46,229][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:48:46,728][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:48:47,227][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:48:47,727][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:48:48,229][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:48:48,729][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:48:49,228][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:48:49,727][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:48:50,226][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:48:50,724][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:48:51,223][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:48:51,721][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:48:52,218][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:48:52,715][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:48:53,213][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:48:53,710][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:48:54,208][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:48:54,704][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:48:55,202][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:48:55,698][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:48:56,196][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:48:56,695][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:48:57,193][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:48:57,691][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:48:58,190][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:48:58,688][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:48:59,186][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:48:59,685][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:49:00,184][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:49:00,683][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:49:01,183][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:49:01,682][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:49:02,181][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:49:02,680][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:49:03,179][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:49:03,680][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21721 tokens. [2026-03-25 22:49:04,323][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 22:49:05,059][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:49:05,062][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:49:05,063][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:49:05,883][__main__][INFO] - Iteration 100 took 1m 15s (10.09% Gen, 88.82% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 22m 14s. Estimated total time: 62h 36m 38s. Time estimates for 10 more iterations: 12m 31s, 100 more iterations: 2h 5m 13s, 500 more iterations: 10h 26m 6s. [2026-03-25 22:49:05,885][__main__][INFO] - Starting iteration 100. [2026-03-25 22:49:06,288][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 1 and human policies 1. [2026-03-25 22:49:06,289][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:49:06,876][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:49:12,239][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:49:12,291][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:49:13,329][__main__][INFO] - Number of regex retries in iteration 100: 3 [2026-03-25 22:49:13,329][__main__][INFO] - agents played in iteration 100 are Alice, Bob [2026-03-25 22:49:14,206][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:49:14,250][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:49:14,251][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:49:14,251][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:49:14,864][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:49:15,320][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:49:15,825][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:49:16,324][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:49:16,822][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:49:17,321][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:49:17,819][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:49:18,319][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:49:18,819][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:49:19,318][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:49:19,817][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:49:20,315][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:49:20,814][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:49:21,313][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:49:21,812][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:49:22,310][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:49:22,809][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:49:23,309][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:49:23,809][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:49:24,307][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:49:24,805][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:49:25,304][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:49:25,803][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:49:26,304][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:49:26,803][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:49:27,306][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:49:27,806][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:49:28,305][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:49:28,803][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:49:29,302][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:49:29,801][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:49:30,299][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:49:30,798][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:49:31,297][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:49:31,796][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:49:32,294][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:49:32,792][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:49:33,290][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:49:33,789][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:49:34,289][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:49:34,788][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:49:35,287][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:49:35,786][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:49:36,284][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:49:36,781][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:49:37,279][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:49:37,778][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:49:38,276][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:49:38,776][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:49:39,276][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:49:39,774][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:49:40,273][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:49:40,771][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:49:41,269][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:49:41,769][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:49:42,269][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:49:42,768][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:49:43,267][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:49:43,766][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:49:44,264][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:49:44,762][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:49:45,260][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:49:45,758][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:49:46,257][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:49:46,756][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:49:47,254][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:49:47,753][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:49:48,254][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:49:48,753][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:49:49,251][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:49:49,749][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:49:50,249][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:49:50,747][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:49:51,246][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:49:51,746][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:49:52,245][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:49:52,743][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:49:53,241][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:49:53,739][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:49:54,238][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:49:54,737][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:49:55,237][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:49:55,736][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:49:56,235][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:49:56,734][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:49:57,233][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:49:57,731][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:49:58,231][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:49:58,730][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:49:59,230][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:49:59,729][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:50:00,228][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:50:00,728][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:50:01,228][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:50:01,729][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:50:02,229][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:50:02,729][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:50:03,228][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:50:03,727][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:50:04,226][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:50:04,726][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:50:05,225][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:50:05,725][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:50:06,224][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:50:06,723][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:50:07,221][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:50:07,720][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:50:08,218][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:50:08,716][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:50:09,215][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:50:09,715][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:50:10,216][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:50:10,715][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:50:11,213][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:50:11,712][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:50:12,212][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:50:12,712][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:50:13,211][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:50:13,710][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:50:14,210][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:50:14,708][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:50:15,206][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:50:15,704][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:50:16,202][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:50:16,700][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:50:17,197][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:50:17,696][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:50:18,194][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:50:18,694][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-25 22:50:19,337][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 22:50:20,074][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:50:20,076][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:50:20,078][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:50:21,291][__main__][INFO] - Iteration 101 took 1m 15s (9.39% Gen, 88.99% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 14m 31s. Estimated total time: 62h 30m 10s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 0s, 500 more iterations: 10h 25m 1s. [2026-03-25 22:50:21,293][__main__][INFO] - Starting iteration 101. [2026-03-25 22:50:21,696][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:50:21,697][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:50:29,439][__main__][INFO] - Number of regex retries in iteration 101: 0 [2026-03-25 22:50:29,440][__main__][INFO] - agents played in iteration 101 are Alice, Bob [2026-03-25 22:50:30,318][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:50:30,363][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:50:30,363][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:50:30,364][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:50:30,971][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:50:31,428][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:50:31,932][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:50:32,432][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:50:32,931][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:50:33,431][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:50:33,931][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:50:34,430][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:50:34,930][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:50:35,430][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:50:35,929][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:50:36,430][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:50:36,930][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:50:37,430][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:50:37,929][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:50:38,429][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:50:38,929][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:50:39,429][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:50:39,930][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:50:40,429][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:50:40,929][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:50:41,428][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:50:41,927][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:50:42,426][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:50:42,925][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:50:43,423][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:50:43,921][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:50:44,419][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:50:44,917][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:50:45,418][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:50:45,919][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:50:46,418][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:50:46,916][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:50:47,414][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:50:47,913][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:50:48,412][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:50:48,911][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:50:49,410][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:50:49,908][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:50:50,407][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:50:50,905][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:50:51,404][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:50:51,903][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:50:52,401][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:50:52,900][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:50:53,399][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:50:53,899][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:50:54,398][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:50:54,897][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:50:55,396][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:50:55,893][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:50:56,387][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:50:56,888][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:50:57,382][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:50:57,877][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:50:58,372][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:50:58,871][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:50:59,369][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:50:59,867][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:51:00,364][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:51:00,862][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:51:01,359][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:51:01,857][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:51:02,355][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:51:02,854][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:51:03,354][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:51:03,853][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:51:04,352][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:51:04,853][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:51:05,353][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:51:05,853][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:51:06,352][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:51:06,852][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:51:07,352][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:51:07,850][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:51:08,349][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:51:08,848][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:51:09,349][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:51:09,849][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:51:10,349][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:51:10,849][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:51:11,348][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:51:11,846][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:51:12,345][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:51:12,845][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:51:13,345][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:51:13,846][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:51:14,345][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:51:14,843][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:51:15,342][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:51:15,840][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:51:16,338][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:51:16,836][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:51:17,335][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:51:17,834][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:51:18,335][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:51:18,835][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:51:19,334][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:51:19,833][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:51:20,331][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:51:20,831][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:51:21,330][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:51:21,828][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:51:22,326][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:51:22,824][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:51:23,323][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:51:23,823][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:51:24,322][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:51:24,822][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:51:25,322][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:51:25,823][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:51:26,322][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:51:26,821][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:51:27,319][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:51:27,818][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:51:28,317][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:51:28,815][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:51:29,315][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:51:29,814][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:51:30,312][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:51:30,811][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:51:31,310][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:51:31,809][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:51:32,309][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:51:32,808][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:51:33,307][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:51:33,806][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:51:34,304][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:51:34,800][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21603 tokens. [2026-03-25 22:51:35,429][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-25 22:51:36,168][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:51:36,170][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:51:36,172][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:51:37,034][__main__][INFO] - Iteration 102 took 1m 15s (10.28% Gen, 88.58% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 60h 30m 0s. Estimated total time: 62h 46m 55s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 33s, 500 more iterations: 10h 27m 49s. [2026-03-25 22:51:37,036][__main__][INFO] - Starting iteration 102. [2026-03-25 22:51:37,437][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:51:37,438][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:51:43,975][__main__][INFO] - Number of regex retries in iteration 102: 0 [2026-03-25 22:51:43,976][__main__][INFO] - agents played in iteration 102 are Alice, Bob [2026-03-25 22:51:44,852][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:51:45,177][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:51:45,177][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:51:45,178][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:51:45,788][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:51:46,246][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:51:46,751][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:51:47,251][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:51:47,751][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:51:48,254][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:51:48,754][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:51:49,255][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:51:49,756][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:51:50,255][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:51:50,753][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:51:51,251][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:51:51,750][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:51:52,248][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:51:52,748][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:51:53,248][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:51:53,747][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:51:54,246][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:51:54,745][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:51:55,243][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:51:55,742][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:51:56,240][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:51:56,739][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:51:57,237][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:51:57,737][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:51:58,233][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:51:58,732][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:51:59,230][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:51:59,726][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:52:00,224][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:52:00,719][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:52:01,216][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:52:01,712][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:52:02,212][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:52:02,711][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:52:03,209][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:52:03,708][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:52:04,207][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:52:04,706][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:52:05,206][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:52:05,707][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:52:06,206][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:52:06,705][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:52:07,203][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:52:07,701][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:52:08,199][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:52:08,698][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:52:09,197][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:52:09,696][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:52:10,195][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:52:10,693][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:52:11,191][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:52:11,689][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:52:12,187][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:52:12,684][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:52:13,184][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:52:13,683][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:52:14,184][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:52:14,684][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:52:15,184][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:52:15,687][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:52:16,188][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:52:16,690][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:52:17,191][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:52:17,691][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:52:18,192][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:52:18,693][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:52:19,193][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:52:19,693][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:52:20,193][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:52:20,694][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:52:21,194][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:52:21,694][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:52:22,191][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:52:22,688][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:52:23,186][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:52:23,683][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:52:24,180][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:52:24,677][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:52:25,176][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:52:25,674][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:52:26,171][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:52:26,669][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:52:27,168][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:52:27,666][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:52:28,165][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:52:28,663][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:52:29,162][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:52:29,660][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:52:30,158][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:52:30,656][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:52:31,154][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:52:31,653][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:52:32,152][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:52:32,651][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:52:33,150][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:52:33,649][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:52:34,146][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:52:34,646][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:52:35,144][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:52:35,642][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:52:36,141][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:52:36,641][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:52:37,140][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:52:37,639][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:52:38,137][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:52:38,635][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:52:39,134][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:52:39,633][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:52:40,132][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:52:40,631][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:52:41,130][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:52:41,629][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:52:42,127][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:52:42,625][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:52:43,123][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:52:43,621][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:52:44,119][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:52:44,617][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:52:45,116][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:52:45,614][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:52:46,115][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:52:46,614][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:52:47,115][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:52:47,615][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:52:48,117][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:52:48,619][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:52:49,119][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:52:49,618][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21665 tokens. [2026-03-25 22:52:50,254][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.50%, ΔTime: 00:01:04 [2026-03-25 22:52:50,990][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:52:50,992][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:52:50,994][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:52:51,638][__main__][INFO] - Iteration 103 took 1m 14s (8.81% Gen, 90.32% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 59h 31m 54s. Estimated total time: 61h 50m 4s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 40s, 500 more iterations: 10h 18m 20s. [2026-03-25 22:52:51,640][__main__][INFO] - Starting iteration 103. [2026-03-25 22:52:52,043][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:52:52,044][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:52:59,346][__main__][INFO] - Number of regex retries in iteration 103: 0 [2026-03-25 22:52:59,347][__main__][INFO] - agents played in iteration 103 are Alice, Bob [2026-03-25 22:53:00,228][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:53:00,272][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:53:00,273][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:53:00,273][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:53:00,878][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:53:01,332][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:53:01,835][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:53:02,335][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:53:02,834][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:53:03,332][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:53:03,830][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:53:04,328][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:53:04,826][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:53:05,323][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:53:05,820][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:53:06,318][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:53:06,815][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:53:07,313][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:53:07,810][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:53:08,307][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:53:08,805][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:53:09,304][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:53:09,802][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:53:10,300][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:53:10,798][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:53:11,296][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:53:11,795][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:53:12,294][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:53:12,792][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:53:13,293][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:53:13,792][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:53:14,292][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:53:14,793][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:53:15,292][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:53:15,792][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:53:16,291][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:53:16,790][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:53:17,288][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:53:17,786][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:53:18,288][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:53:18,789][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:53:19,287][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:53:19,786][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:53:20,284][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:53:20,782][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:53:21,284][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:53:21,784][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:53:22,283][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:53:22,782][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:53:23,281][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:53:23,780][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:53:24,280][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:53:24,779][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:53:25,307][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:53:25,809][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:53:26,311][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:53:26,813][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:53:27,314][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:53:27,816][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:53:28,318][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:53:28,819][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:53:29,318][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:53:29,816][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:53:30,315][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:53:30,813][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:53:31,312][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:53:31,812][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:53:32,311][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:53:32,810][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:53:33,307][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:53:33,805][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:53:34,303][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:53:34,800][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:53:35,298][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:53:35,796][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:53:36,295][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:53:36,793][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:53:37,291][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:53:37,788][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:53:38,285][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:53:38,782][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:53:39,278][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:53:39,776][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:53:40,274][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:53:40,771][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:53:41,269][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:53:41,763][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:53:42,262][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:53:42,760][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:53:43,259][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:53:43,757][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:53:44,256][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:53:44,754][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:53:45,253][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:53:45,755][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:53:46,255][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:53:46,754][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:53:47,253][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:53:47,753][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:53:48,254][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:53:48,754][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:53:49,253][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:53:49,751][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:53:50,250][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:53:50,747][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:53:51,245][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:53:51,744][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:53:52,243][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:53:52,741][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:53:53,241][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:53:53,740][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:53:54,240][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:53:54,739][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:53:55,239][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:53:55,739][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:53:56,240][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:53:56,741][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:53:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:53:57,744][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:53:58,245][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:53:58,748][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:53:59,250][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:53:59,752][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:54:00,254][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:54:00,756][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:54:01,254][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:54:01,752][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:54:02,250][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:54:02,749][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:54:03,247][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:54:03,746][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:54:04,243][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:54:04,741][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21701 tokens. [2026-03-25 22:54:05,368][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-25 22:54:06,094][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:54:06,097][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:54:06,098][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:54:06,736][__main__][INFO] - Iteration 104 took 1m 14s (9.78% Gen, 89.37% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 55m 16s. Estimated total time: 62h 14m 41s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 29s, 500 more iterations: 10h 22m 26s. [2026-03-25 22:54:06,739][__main__][INFO] - Starting iteration 104. [2026-03-25 22:54:07,140][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:54:07,141][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:54:14,109][__main__][INFO] - Number of regex retries in iteration 104: 0 [2026-03-25 22:54:14,110][__main__][INFO] - agents played in iteration 104 are Alice, Bob [2026-03-25 22:54:14,988][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:54:15,032][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:54:15,033][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:54:15,034][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:54:15,636][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:54:16,093][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:54:16,597][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:54:17,096][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:54:17,597][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:54:18,097][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:54:18,600][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:54:19,099][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:54:19,599][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:54:20,098][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:54:20,597][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:54:21,097][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:54:21,596][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:54:22,095][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:54:22,594][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:54:23,093][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:54:23,593][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:54:24,092][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:54:24,590][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:54:25,089][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:54:25,588][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:54:26,086][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:54:26,584][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:54:27,082][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:54:27,579][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:54:28,078][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:54:28,577][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:54:29,076][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:54:29,575][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:54:30,074][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:54:30,573][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:54:31,072][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:54:31,571][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:54:32,070][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:54:32,568][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:54:33,066][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:54:33,563][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:54:34,061][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:54:34,558][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:54:35,055][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:54:35,554][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:54:36,053][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:54:36,552][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:54:37,053][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:54:37,552][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:54:38,051][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:54:38,549][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:54:39,048][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:54:39,546][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:54:40,044][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:54:40,544][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:54:41,042][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:54:41,541][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:54:42,040][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:54:42,539][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:54:43,039][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:54:43,538][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:54:44,037][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:54:44,537][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:54:45,036][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:54:45,536][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:54:46,036][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:54:46,535][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:54:47,035][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:54:47,536][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:54:48,037][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:54:48,539][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:54:49,040][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:54:49,540][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:54:50,039][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:54:50,539][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:54:51,040][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:54:51,540][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:54:52,040][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:54:52,538][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:54:53,038][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:54:53,537][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:54:54,036][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:54:54,535][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:54:55,035][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:54:55,534][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:54:56,033][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:54:56,531][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:54:57,031][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:54:57,531][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:54:58,030][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:54:58,529][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:54:59,029][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:54:59,527][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:55:00,026][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:55:00,524][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:55:01,022][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:55:01,520][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:55:02,018][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:55:02,516][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:55:03,016][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:55:03,516][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:55:04,014][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:55:04,511][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:55:05,009][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:55:05,507][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:55:06,003][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:55:06,500][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:55:06,998][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:55:07,495][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:55:07,995][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:55:08,493][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:55:08,991][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:55:09,490][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:55:09,989][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:55:10,487][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:55:10,985][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:55:11,483][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:55:11,981][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:55:12,481][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:55:12,980][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:55:13,478][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:55:13,977][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:55:14,476][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:55:14,975][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:55:15,477][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:55:15,979][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:55:16,478][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:55:16,977][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:55:17,476][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:55:19,143][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:55:19,644][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:55:20,143][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:55:20,642][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21660 tokens. [2026-03-25 22:55:21,266][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:05 [2026-03-25 22:55:22,002][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:55:22,004][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:55:22,005][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:55:22,823][__main__][INFO] - Iteration 105 took 1m 15s (9.21% Gen, 89.71% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 60h 43m 27s. Estimated total time: 63h 4m 8s. Time estimates for 10 more iterations: 12m 36s, 100 more iterations: 2h 6m 8s, 500 more iterations: 10h 30m 41s. [2026-03-25 22:55:22,825][__main__][INFO] - Starting iteration 105. [2026-03-25 22:55:23,223][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:55:23,223][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:55:29,747][__main__][INFO] - Number of regex retries in iteration 105: 0 [2026-03-25 22:55:29,748][__main__][INFO] - agents played in iteration 105 are Alice, Bob [2026-03-25 22:55:30,625][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:55:30,670][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:55:30,670][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:55:30,671][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:55:31,275][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:55:31,729][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:55:32,233][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:55:32,731][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:55:33,230][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:55:33,730][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:55:34,229][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:55:34,729][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:55:35,227][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:55:35,728][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:55:36,227][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:55:36,726][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:55:37,224][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:55:37,723][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:55:38,222][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:55:38,721][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:55:39,220][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:55:39,720][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:55:40,218][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:55:40,718][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:55:41,217][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:55:41,716][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:55:42,217][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:55:42,716][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:55:43,215][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:55:43,714][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:55:44,214][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:55:44,713][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:55:45,212][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:55:45,711][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:55:46,210][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:55:46,710][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:55:47,210][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:55:47,708][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:55:48,209][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:55:48,710][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:55:49,210][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:55:49,712][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:55:50,211][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:55:50,710][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:55:51,209][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:55:51,707][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:55:52,204][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:55:52,701][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:55:53,199][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:55:53,697][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:55:54,195][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:55:54,693][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:55:55,190][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:55:55,691][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:55:56,192][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:55:56,692][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:55:57,192][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:55:57,692][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:55:58,191][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:55:58,691][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:55:59,191][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:55:59,688][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:56:00,187][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:56:00,684][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:56:01,181][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:56:01,678][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:56:02,176][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:56:02,674][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:56:03,171][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:56:03,668][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:56:04,166][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:56:04,663][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:56:05,161][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:56:05,659][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:56:06,158][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:56:06,656][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:56:07,155][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:56:07,654][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:56:08,152][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:56:08,652][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:56:09,150][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:56:09,649][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:56:10,151][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:56:10,651][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:56:11,150][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:56:11,649][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:56:12,148][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:56:12,647][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:56:13,146][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:56:13,646][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:56:14,145][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:56:14,645][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:56:15,145][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:56:15,644][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:56:16,142][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:56:16,641][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:56:17,140][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:56:17,640][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:56:18,140][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:56:18,641][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:56:19,141][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:56:19,640][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:56:20,138][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:56:20,640][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:56:21,140][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:56:21,640][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:56:22,139][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:56:22,637][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:56:23,136][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:56:23,634][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:56:24,132][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:56:24,630][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:56:25,128][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:56:25,626][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:56:26,124][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:56:26,620][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:56:27,117][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:56:27,616][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:56:28,116][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:56:28,616][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:56:29,116][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:56:29,617][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:56:30,117][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:56:30,616][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:56:31,116][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:56:31,614][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:56:32,113][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:56:32,611][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:56:33,109][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:56:33,607][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:56:34,104][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:56:34,602][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:56:35,099][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-25 22:56:35,739][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 22:56:36,536][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:56:36,538][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:56:36,540][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:56:37,284][__main__][INFO] - Iteration 106 took 1m 14s (8.81% Gen, 90.18% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 21m 10s. Estimated total time: 61h 43m 5s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 26s, 500 more iterations: 10h 17m 10s. [2026-03-25 22:56:37,286][__main__][INFO] - Starting iteration 106. [2026-03-25 22:56:37,686][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:56:37,687][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:56:44,346][__main__][INFO] - Number of regex retries in iteration 106: 0 [2026-03-25 22:56:44,348][__main__][INFO] - agents played in iteration 106 are Alice, Bob [2026-03-25 22:56:45,232][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:56:45,277][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:56:45,277][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:56:45,278][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:56:45,884][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:56:46,342][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:56:46,843][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:56:47,341][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:56:47,839][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:56:48,337][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:56:48,837][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:56:49,335][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:56:49,833][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:56:50,332][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:56:50,831][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:56:51,330][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:56:51,829][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:56:52,328][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:56:52,826][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:56:53,324][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:56:53,822][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:56:54,320][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:56:54,819][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:56:55,318][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:56:55,816][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:56:56,315][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:56:56,808][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:56:57,307][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:56:57,805][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:56:58,305][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:56:58,804][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:56:59,303][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:56:59,802][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:57:00,301][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:57:00,800][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:57:01,301][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:57:01,801][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:57:02,299][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:57:02,798][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:57:03,297][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:57:03,796][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:57:04,296][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:57:04,794][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:57:05,294][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:57:05,795][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:57:06,291][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:57:06,787][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:57:07,282][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:57:07,779][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:57:08,275][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:57:08,769][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:57:09,264][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:57:09,759][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:57:10,258][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:57:10,758][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:57:11,258][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:57:11,758][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:57:12,258][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:57:12,757][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:57:13,257][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:57:13,757][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:57:14,258][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:57:14,757][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:57:15,256][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:57:15,755][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:57:16,255][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:57:16,755][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:57:17,255][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:57:17,754][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:57:18,256][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:57:18,754][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:57:19,253][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:57:19,751][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:57:20,250][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:57:20,750][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:57:21,247][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:57:21,987][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:57:22,486][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:57:22,985][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:57:23,484][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:57:23,982][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:57:24,481][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:57:24,980][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:57:25,479][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:57:25,979][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:57:26,478][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:57:26,976][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:57:27,475][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:57:27,974][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:57:28,472][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:57:28,971][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:57:29,469][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:57:29,969][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:57:30,469][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:57:30,969][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:57:31,467][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:57:31,968][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:57:32,468][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:57:32,969][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:57:33,469][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:57:33,968][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:57:34,468][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:57:34,966][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:57:35,466][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:57:35,965][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:57:36,465][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:57:36,965][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:57:37,464][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:57:37,963][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:57:38,458][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:57:38,953][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:57:39,448][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:57:39,943][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:57:40,438][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:57:40,933][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:57:41,429][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:57:41,924][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:57:42,423][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:57:42,923][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:57:43,423][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:57:43,924][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:57:44,422][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:57:44,922][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:57:45,420][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:57:45,920][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:57:46,419][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:57:46,918][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:57:47,417][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:57:47,918][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:57:48,419][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:57:48,918][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:57:49,419][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:57:49,919][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21689 tokens. [2026-03-25 22:57:50,563][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 22:57:51,308][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:57:51,310][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:57:51,312][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:57:51,970][__main__][INFO] - Iteration 107 took 1m 14s (8.97% Gen, 90.15% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 31m 2s. Estimated total time: 61h 54m 12s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 48s, 500 more iterations: 10h 19m 2s. [2026-03-25 22:57:51,972][__main__][INFO] - Starting iteration 107. [2026-03-25 22:57:52,376][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:57:52,377][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:57:52,963][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:57:59,214][mllm.models.large_language_model_local][WARNING] - Response Given the symmetric nature of the per-item values and the fact that both you and Alice have the same values for hats and books, and balls, the best strategy is to mirror Alice's proposal to ensure equitable distribution and maximize points based on the highest valued items (hats and books). Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:58:00,416][__main__][INFO] - Number of regex retries in iteration 107: 2 [2026-03-25 22:58:00,417][__main__][INFO] - agents played in iteration 107 are Alice, Bob [2026-03-25 22:58:01,294][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:58:01,339][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:58:01,339][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:58:01,340][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:58:01,939][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:58:02,395][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:58:02,898][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:58:03,397][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:58:03,896][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:58:04,396][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:58:04,894][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:58:05,393][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:58:05,893][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:58:06,388][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:58:06,884][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:58:07,380][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:58:07,876][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:58:08,373][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:58:08,870][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:58:09,366][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:58:09,863][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:58:10,361][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:58:10,859][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:58:11,358][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:58:11,857][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:58:12,355][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:58:12,854][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:58:13,353][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:58:13,852][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:58:14,350][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:58:14,848][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:58:15,348][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:58:15,848][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:58:16,347][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:58:16,846][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:58:17,344][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:58:17,842][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:58:18,345][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:58:18,846][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:58:19,345][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:58:19,845][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:58:20,344][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:58:20,843][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:58:21,343][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:58:21,844][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:58:22,343][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:58:22,841][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:58:23,339][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:58:23,840][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:58:24,338][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:58:24,836][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:58:25,334][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:58:25,832][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:58:26,331][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:58:26,830][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:58:27,331][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:58:27,831][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:58:28,332][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:58:28,832][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:58:29,332][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:58:29,832][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:58:30,332][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:58:30,832][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:58:31,332][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:58:31,831][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:58:32,330][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:58:32,829][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:58:33,329][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:58:33,829][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:58:34,329][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:58:34,831][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:58:35,331][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:58:35,831][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:58:36,329][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:58:36,828][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:58:37,327][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:58:37,826][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:58:38,321][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:58:38,818][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:58:39,316][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:58:39,814][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:58:40,310][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:58:40,807][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:58:41,303][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:58:41,798][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:58:42,297][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:58:42,795][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:58:43,294][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:58:43,793][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:58:44,293][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:58:44,791][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:58:45,291][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 22:58:45,788][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 22:58:46,286][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 22:58:46,784][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 22:58:47,281][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 22:58:47,779][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 22:58:48,278][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 22:58:48,777][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 22:58:49,276][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 22:58:49,773][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 22:58:50,271][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 22:58:50,769][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 22:58:51,268][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 22:58:51,766][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 22:58:52,265][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 22:58:52,763][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 22:58:53,263][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 22:58:53,762][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 22:58:54,259][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 22:58:54,759][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 22:58:55,258][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 22:58:55,757][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 22:58:56,255][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 22:58:56,754][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 22:58:57,253][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 22:58:57,751][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 22:58:58,249][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 22:58:58,748][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 22:58:59,247][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 22:58:59,747][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 22:59:00,246][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 22:59:00,746][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 22:59:01,246][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 22:59:01,745][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 22:59:02,244][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 22:59:02,741][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 22:59:03,240][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 22:59:03,738][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 22:59:04,237][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 22:59:04,736][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 22:59:05,235][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 22:59:05,734][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21690 tokens. [2026-03-25 22:59:06,361][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-25 22:59:07,096][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 22:59:07,098][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 22:59:07,100][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 22:59:07,745][__main__][INFO] - Iteration 108 took 1m 15s (10.67% Gen, 88.48% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 60h 24m 1s. Estimated total time: 62h 48m 27s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 36s, 500 more iterations: 10h 28m 4s. [2026-03-25 22:59:07,747][__main__][INFO] - Starting iteration 108. [2026-03-25 22:59:08,145][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 22:59:08,146][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 22:59:10,407][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 22:59:14,693][__main__][INFO] - Number of regex retries in iteration 108: 1 [2026-03-25 22:59:14,694][__main__][INFO] - agents played in iteration 108 are Alice, Bob [2026-03-25 22:59:15,573][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:59:15,617][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 22:59:15,618][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 22:59:15,618][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 22:59:16,227][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 22:59:16,684][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 22:59:17,189][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 22:59:17,689][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 22:59:18,192][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 22:59:18,693][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 22:59:19,194][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 22:59:19,694][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 22:59:20,194][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 22:59:20,693][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 22:59:21,193][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 22:59:21,692][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 22:59:22,190][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 22:59:22,689][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 22:59:23,188][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 22:59:23,688][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 22:59:24,188][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 22:59:24,685][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 22:59:25,183][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 22:59:25,681][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 22:59:26,178][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 22:59:26,676][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 22:59:27,174][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 22:59:27,673][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 22:59:28,173][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 22:59:28,673][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 22:59:29,172][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 22:59:29,669][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 22:59:30,168][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 22:59:30,666][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 22:59:31,164][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 22:59:31,663][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 22:59:32,160][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 22:59:32,658][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 22:59:33,156][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 22:59:33,655][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 22:59:34,152][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 22:59:34,650][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 22:59:35,147][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 22:59:35,646][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 22:59:36,143][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 22:59:36,643][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 22:59:37,141][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 22:59:37,640][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 22:59:38,139][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 22:59:38,639][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 22:59:39,138][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 22:59:39,636][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 22:59:40,135][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 22:59:40,635][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 22:59:41,133][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 22:59:41,632][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 22:59:42,131][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 22:59:42,628][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 22:59:43,126][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 22:59:43,623][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 22:59:44,120][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 22:59:44,618][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 22:59:45,116][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 22:59:45,617][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 22:59:46,117][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 22:59:46,615][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 22:59:47,113][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 22:59:47,611][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 22:59:48,109][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 22:59:48,612][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 22:59:49,112][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 22:59:49,612][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 22:59:50,112][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 22:59:50,611][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 22:59:51,110][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 22:59:51,610][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 22:59:52,110][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 22:59:52,610][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 22:59:53,109][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 22:59:53,607][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 22:59:54,106][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 22:59:54,604][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 22:59:55,102][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 22:59:55,600][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 22:59:56,099][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 22:59:56,598][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 22:59:57,095][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 22:59:57,593][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 22:59:58,091][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 22:59:58,588][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 22:59:59,086][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 22:59:59,583][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:00:00,082][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:00:00,580][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:00:01,078][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:00:01,576][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:00:02,074][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:00:02,572][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:00:03,069][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:00:03,566][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:00:04,063][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:00:04,560][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:00:05,058][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:00:05,557][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:00:06,055][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:00:06,554][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:00:07,052][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:00:07,549][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:00:08,046][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:00:08,544][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:00:09,043][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:00:09,542][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:00:10,042][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:00:10,542][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:00:11,041][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:00:11,539][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:00:12,038][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:00:12,536][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:00:13,034][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:00:13,533][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:00:14,032][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:00:14,530][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:00:15,028][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:00:15,526][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:00:16,023][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:00:16,520][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:00:17,018][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:00:17,516][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:00:18,015][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:00:18,514][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:00:19,012][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:00:19,509][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:00:20,006][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-25 23:00:20,634][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:00:21,381][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:00:21,383][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:00:21,385][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:00:22,052][__main__][INFO] - Iteration 109 took 1m 13s (8.86% Gen, 90.24% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 9m 40s. Estimated total time: 61h 35m 20s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 10s, 500 more iterations: 10h 15m 53s. [2026-03-25 23:00:22,054][__main__][INFO] - Starting iteration 109. [2026-03-25 23:00:22,454][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:00:22,455][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:00:29,021][__main__][INFO] - Number of regex retries in iteration 109: 0 [2026-03-25 23:00:29,022][__main__][INFO] - agents played in iteration 109 are Alice, Bob [2026-03-25 23:00:29,909][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:00:29,954][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:00:29,955][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:00:29,955][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:00:30,574][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:00:31,031][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:00:31,535][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:00:32,035][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:00:32,535][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:00:33,035][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:00:33,535][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:00:34,035][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:00:34,535][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:00:35,037][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:00:35,538][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:00:36,038][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:00:36,539][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:00:37,040][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:00:37,541][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:00:38,042][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:00:38,542][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:00:39,041][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:00:39,540][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:00:40,039][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:00:40,538][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:00:41,038][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:00:41,538][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:00:42,037][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:00:42,536][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:00:43,036][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:00:43,537][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:00:44,037][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:00:44,536][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:00:45,036][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:00:45,536][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:00:46,037][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:00:46,537][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:00:47,036][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:00:47,536][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:00:48,035][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:00:48,536][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:00:49,035][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:00:49,534][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:00:50,033][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:00:50,531][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:00:51,030][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:00:51,530][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:00:52,028][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:00:52,527][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:00:53,026][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:00:53,525][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:00:54,023][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:00:54,522][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:00:55,020][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:00:55,519][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:00:56,017][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:00:56,515][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:00:57,013][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:00:57,514][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:00:58,011][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:00:58,508][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:00:59,005][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:00:59,503][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:01:00,000][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:01:00,498][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:01:00,996][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:01:01,494][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:01:01,992][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:01:02,490][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:01:02,988][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:01:03,487][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:01:03,984][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:01:04,482][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:01:04,980][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:01:05,479][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:01:05,978][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:01:06,476][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:01:06,975][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:01:07,474][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:01:07,974][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:01:08,473][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:01:08,974][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:01:09,474][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:01:09,974][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:01:10,472][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:01:10,970][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:01:11,467][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:01:11,964][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:01:12,462][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:01:12,961][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:01:13,460][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:01:13,958][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:01:14,456][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:01:14,956][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:01:15,458][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:01:15,960][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:01:16,460][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:01:16,960][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:01:17,459][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:01:17,961][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:01:18,463][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:01:18,961][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:01:19,459][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:01:19,958][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:01:20,455][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:01:20,954][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:01:21,453][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:01:21,951][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:01:22,449][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:01:22,948][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:01:23,446][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:01:23,944][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:01:24,442][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:01:24,941][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:01:25,442][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:01:25,940][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:01:26,440][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:01:26,939][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:01:27,438][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:01:27,937][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:01:28,437][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:01:28,936][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:01:29,435][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:01:29,934][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:01:30,435][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:01:30,935][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:01:31,434][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:01:31,933][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:01:32,432][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:01:32,930][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:01:33,429][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:01:33,926][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:01:34,424][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21745 tokens. [2026-03-25 23:01:35,053][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 23:01:35,839][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:01:35,842][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:01:35,843][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:01:36,664][__main__][INFO] - Iteration 110 took 1m 14s (8.85% Gen, 90.04% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 23m 36s. Estimated total time: 61h 50m 30s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 25s. [2026-03-25 23:01:36,666][__main__][INFO] - Starting iteration 110. [2026-03-25 23:01:37,072][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:01:37,073][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:01:38,223][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:01:38,722][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:01:43,460][__main__][INFO] - Number of regex retries in iteration 110: 2 [2026-03-25 23:01:43,461][__main__][INFO] - agents played in iteration 110 are Alice, Bob [2026-03-25 23:01:44,340][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:01:44,661][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:01:44,662][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:01:44,663][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:01:45,263][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:01:45,717][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:01:46,221][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:01:46,720][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:01:47,219][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:01:47,717][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:01:48,218][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:01:48,718][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:01:49,218][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:01:49,716][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:01:50,215][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:01:50,714][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:01:51,213][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:01:51,712][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:01:52,211][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:01:52,708][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:01:53,208][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:01:53,705][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:01:54,201][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:01:54,699][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:01:55,197][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:01:55,694][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:01:56,192][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:01:56,689][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:01:57,186][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:01:57,685][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:01:58,183][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:01:58,682][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:01:59,181][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:01:59,680][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:02:00,178][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:02:00,677][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:02:01,176][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:02:01,671][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:02:02,166][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:02:02,662][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:02:03,157][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:02:03,652][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:02:04,147][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:02:04,643][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:02:05,137][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:02:05,635][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:02:06,130][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:02:06,629][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:02:07,124][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:02:07,622][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:02:08,120][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:02:08,619][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:02:09,118][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:02:09,618][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:02:10,117][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:02:10,616][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:02:11,116][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:02:11,614][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:02:12,114][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:02:12,614][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:02:13,113][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:02:13,611][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:02:14,108][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:02:14,607][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:02:15,105][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:02:15,603][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:02:16,101][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:02:16,601][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:02:17,099][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:02:17,596][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:02:18,095][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:02:18,596][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:02:19,094][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:02:19,592][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:02:20,090][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:02:20,589][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:02:21,086][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:02:21,585][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:02:22,083][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:02:22,581][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:02:23,080][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:02:23,578][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:02:24,076][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:02:24,575][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:02:25,074][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:02:25,572][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:02:26,070][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:02:26,568][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:02:27,065][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:02:27,564][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:02:28,061][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:02:28,558][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:02:29,056][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:02:29,554][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:02:30,054][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:02:30,552][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:02:31,051][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:02:31,549][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:02:32,047][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:02:32,545][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:02:33,044][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:02:33,539][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:02:34,035][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:02:34,531][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:02:35,026][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:02:35,521][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:02:36,015][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:02:36,510][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:02:37,004][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:02:37,502][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:02:37,999][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:02:38,497][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:02:38,995][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:02:39,493][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:02:39,991][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:02:40,489][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:02:40,987][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:02:41,486][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:02:41,985][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:02:42,483][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:02:42,982][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:02:43,481][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:02:43,980][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:02:44,479][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:02:44,979][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:02:45,480][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:02:45,980][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:02:46,479][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:02:46,976][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:02:47,474][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:02:47,973][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:02:48,473][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:02:48,971][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21717 tokens. [2026-03-25 23:02:49,607][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 23:02:50,341][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:02:50,343][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:02:50,345][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:02:51,036][__main__][INFO] - Iteration 111 took 1m 13s (8.64% Gen, 90.43% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 10m 4s. Estimated total time: 61h 38m 13s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 16s, 500 more iterations: 10h 16m 22s. [2026-03-25 23:02:51,038][__main__][INFO] - Starting iteration 111. [2026-03-25 23:02:51,438][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:02:51,439][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:02:52,034][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:02:58,899][__main__][INFO] - Number of regex retries in iteration 111: 1 [2026-03-25 23:02:58,900][__main__][INFO] - agents played in iteration 111 are Alice, Bob [2026-03-25 23:02:59,780][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:02:59,824][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:02:59,825][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:02:59,826][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:03:00,432][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:03:00,885][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:03:01,386][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:03:01,883][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:03:02,380][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:03:02,878][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:03:03,376][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:03:03,872][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:03:04,369][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:03:04,869][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:03:05,367][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:03:05,867][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:03:06,365][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:03:06,863][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:03:07,362][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:03:07,861][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:03:08,359][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:03:08,860][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:03:09,359][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:03:09,857][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:03:10,357][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:03:10,857][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:03:11,356][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:03:11,855][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:03:12,355][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:03:12,852][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:03:13,348][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:03:13,844][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:03:14,342][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:03:14,840][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:03:15,337][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:03:15,835][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:03:16,333][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:03:16,830][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:03:17,327][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:03:17,824][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:03:18,324][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:03:18,822][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:03:19,320][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:03:19,817][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:03:20,316][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:03:20,814][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:03:21,312][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:03:21,810][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:03:22,309][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:03:22,807][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:03:23,305][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:03:23,803][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:03:24,302][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:03:24,800][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:03:25,298][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:03:25,796][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:03:26,294][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:03:26,793][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:03:27,292][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:03:27,791][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:03:28,290][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:03:28,791][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:03:29,290][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:03:29,788][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:03:30,287][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:03:30,786][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:03:31,287][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:03:31,786][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:03:32,286][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:03:32,784][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:03:33,281][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:03:33,779][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:03:34,278][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:03:34,777][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:03:35,275][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:03:35,774][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:03:36,273][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:03:36,773][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:03:37,271][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:03:37,769][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:03:38,268][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:03:38,766][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:03:39,264][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:03:39,762][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:03:40,261][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:03:40,761][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:03:41,260][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:03:41,762][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:03:42,261][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:03:42,761][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:03:43,260][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:03:43,759][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:03:44,260][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:03:44,757][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:03:45,255][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:03:45,752][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:03:46,250][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:03:46,748][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:03:47,246][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:03:47,743][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:03:48,244][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:03:48,742][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:03:49,240][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:03:49,738][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:03:50,236][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:03:50,734][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:03:51,232][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:03:51,731][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:03:52,230][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:03:52,728][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:03:53,226][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:03:53,724][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:03:54,222][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:03:54,720][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:03:55,220][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:03:55,720][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:03:56,220][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:03:56,719][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:03:57,217][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:03:57,715][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:03:58,214][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:03:58,712][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:03:59,213][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:03:59,712][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:04:00,211][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:04:00,710][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:04:01,209][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:04:01,708][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:04:02,207][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:04:02,704][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:04:03,204][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:04:03,703][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:04:04,203][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21728 tokens. [2026-03-25 23:04:04,839][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 23:04:05,571][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:04:05,574][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:04:05,575][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:04:06,420][__main__][INFO] - Iteration 112 took 1m 14s (9.95% Gen, 88.92% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 59m 42s. Estimated total time: 62h 29m 6s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 58s, 500 more iterations: 10h 24m 51s. [2026-03-25 23:04:06,422][__main__][INFO] - Starting iteration 112. [2026-03-25 23:04:06,828][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:04:06,829][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:04:08,517][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:04:13,699][__main__][INFO] - Number of regex retries in iteration 112: 1 [2026-03-25 23:04:13,700][__main__][INFO] - agents played in iteration 112 are Alice, Bob [2026-03-25 23:04:14,577][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:04:14,621][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:04:14,622][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:04:14,622][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:04:15,227][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:04:15,685][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:04:16,190][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:04:16,691][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:04:17,190][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:04:17,689][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:04:18,191][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:04:18,691][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:04:19,192][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:04:19,690][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:04:20,189][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:04:20,687][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:04:21,185][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:04:21,684][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:04:22,183][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:04:22,684][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:04:23,182][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:04:23,680][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:04:24,177][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:04:24,674][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:04:25,172][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:04:25,669][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:04:26,167][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:04:26,666][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:04:27,165][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:04:27,664][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:04:28,163][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:04:28,662][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:04:29,162][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:04:29,660][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:04:30,159][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:04:30,657][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:04:31,156][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:04:31,654][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:04:32,152][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:04:32,650][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:04:33,148][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:04:33,646][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:04:34,144][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:04:34,643][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:04:35,142][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:04:35,640][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:04:36,137][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:04:36,636][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:04:37,134][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:04:37,634][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:04:38,133][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:04:38,632][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:04:39,130][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:04:39,628][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:04:40,126][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:04:40,624][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:04:41,122][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:04:41,620][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:04:42,119][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:04:42,616][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:04:43,115][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:04:43,614][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:04:44,112][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:04:44,610][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:04:45,108][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:04:45,605][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:04:46,103][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:04:46,601][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:04:47,099][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:04:47,599][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:04:48,098][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:04:48,600][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:04:49,100][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:04:49,599][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:04:50,098][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:04:50,598][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:04:51,097][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:04:51,596][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:04:52,095][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:04:52,593][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:04:53,092][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:04:53,590][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:04:54,089][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:04:54,588][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:04:55,087][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:04:55,584][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:04:56,082][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:04:56,579][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:04:57,076][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:04:57,574][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:04:58,071][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:04:58,567][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:04:59,065][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:04:59,562][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:05:00,060][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:05:00,559][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:05:01,057][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:05:01,555][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:05:02,053][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:05:02,550][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:05:03,048][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:05:03,546][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:05:04,045][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:05:04,543][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:05:05,041][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:05:05,540][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:05:06,038][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:05:06,537][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:05:07,035][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:05:07,535][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:05:08,034][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:05:08,533][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:05:09,032][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:05:09,529][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:05:10,026][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:05:10,523][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:05:11,021][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:05:11,519][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:05:12,018][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:05:12,516][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:05:13,014][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:05:13,512][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:05:14,010][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:05:14,508][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:05:15,006][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:05:15,504][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:05:16,002][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:05:16,500][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:05:16,999][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:05:17,497][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:05:17,996][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:05:18,497][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:05:18,995][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-25 23:05:19,619][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-25 23:05:20,350][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:05:20,352][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:05:20,354][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:05:20,999][__main__][INFO] - Iteration 113 took 1m 14s (9.26% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 17m 56s. Estimated total time: 61h 48m 35s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 37s, 500 more iterations: 10h 18m 5s. [2026-03-25 23:05:21,001][__main__][INFO] - Starting iteration 113. [2026-03-25 23:05:21,406][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:05:21,407][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:05:22,008][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:05:22,009][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:05:28,630][__main__][INFO] - Number of regex retries in iteration 113: 2 [2026-03-25 23:05:28,631][__main__][INFO] - agents played in iteration 113 are Alice, Bob [2026-03-25 23:05:29,509][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:05:29,553][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:05:29,554][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:05:29,555][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:05:30,155][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:05:30,609][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:05:31,112][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:05:31,609][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:05:32,105][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:05:32,605][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:05:33,103][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:05:33,601][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:05:34,098][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:05:34,599][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:05:35,098][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:05:35,597][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:05:36,096][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:05:36,596][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:05:37,095][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:05:37,594][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:05:38,092][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:05:38,593][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:05:39,092][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:05:39,593][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:05:40,094][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:05:40,593][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:05:41,093][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:05:41,592][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:05:42,091][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:05:42,588][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:05:43,086][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:05:43,583][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:05:44,082][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:05:44,580][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:05:45,077][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:05:45,577][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:05:46,077][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:05:46,599][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:05:47,099][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:05:47,600][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:05:48,100][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:05:48,601][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:05:49,100][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:05:49,601][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:05:50,101][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:05:50,600][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:05:51,100][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:05:51,598][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:05:52,097][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:05:52,595][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:05:53,095][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:05:53,595][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:05:54,095][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:05:54,592][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:05:55,091][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:05:55,588][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:05:56,086][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:05:56,584][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:05:57,082][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:05:57,580][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:05:58,078][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:05:58,577][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:05:59,075][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:05:59,573][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:06:00,071][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:06:00,569][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:06:01,067][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:06:01,565][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:06:02,063][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:06:02,562][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:06:03,060][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:06:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:06:04,055][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:06:04,552][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:06:05,050][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:06:05,546][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:06:06,044][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:06:06,544][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:06:07,044][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:06:07,544][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:06:08,043][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:06:08,542][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:06:09,041][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:06:09,540][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:06:10,040][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:06:10,541][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:06:11,041][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:06:11,540][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:06:12,040][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:06:12,539][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:06:13,038][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:06:13,538][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:06:14,037][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:06:14,536][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:06:15,034][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:06:15,531][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:06:16,028][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:06:16,525][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:06:17,022][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:06:17,519][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:06:18,018][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:06:18,519][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:06:19,019][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:06:19,518][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:06:20,016][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:06:20,515][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:06:21,015][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:06:21,515][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:06:22,016][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:06:22,516][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:06:23,015][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:06:23,514][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:06:24,013][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:06:24,513][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:06:25,012][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:06:25,512][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:06:26,012][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:06:26,510][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:06:27,007][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:06:27,503][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:06:28,000][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:06:28,498][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:06:28,995][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:06:29,491][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:06:29,988][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:06:30,487][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:06:30,985][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:06:31,483][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:06:31,981][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:06:32,480][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:06:32,978][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:06:33,477][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:06:33,974][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21647 tokens. [2026-03-25 23:06:34,605][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:06:35,347][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:06:35,349][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:06:35,351][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:06:36,200][__main__][INFO] - Iteration 114 took 1m 14s (9.66% Gen, 89.20% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 47m 50s. Estimated total time: 62h 19m 44s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 39s, 500 more iterations: 10h 23m 17s. [2026-03-25 23:06:36,203][__main__][INFO] - Starting iteration 114. [2026-03-25 23:06:36,604][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:06:36,605][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:06:43,517][__main__][INFO] - Number of regex retries in iteration 114: 0 [2026-03-25 23:06:43,518][__main__][INFO] - agents played in iteration 114 are Alice, Bob [2026-03-25 23:06:44,400][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:06:44,444][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:06:44,445][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:06:44,445][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:06:45,049][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:06:45,505][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:06:46,007][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:06:46,505][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:06:47,003][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:06:47,502][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:06:48,002][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:06:48,502][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:06:49,001][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:06:49,500][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:06:49,999][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:06:50,497][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:06:50,997][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:06:51,496][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:06:51,995][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:06:52,494][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:06:52,992][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:06:53,491][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:06:53,988][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:06:54,486][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:06:54,984][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:06:55,482][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:06:55,981][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:06:56,479][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:06:56,977][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:06:57,475][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:06:57,973][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:06:58,470][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:06:58,970][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:06:59,469][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:06:59,967][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:07:00,466][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:07:00,964][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:07:01,465][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:07:01,965][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:07:02,465][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:07:02,965][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:07:03,464][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:07:03,965][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:07:04,465][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:07:04,964][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:07:05,461][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:07:05,957][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:07:06,454][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:07:06,953][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:07:07,450][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:07:07,947][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:07:08,445][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:07:08,942][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:07:09,441][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:07:09,941][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:07:10,442][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:07:10,941][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:07:11,442][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:07:11,942][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:07:12,442][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:07:12,942][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:07:13,445][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:07:13,947][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:07:14,449][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:07:14,951][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:07:15,452][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:07:15,954][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:07:16,456][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:07:16,960][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:07:17,460][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:07:17,960][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:07:18,461][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:07:18,961][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:07:19,459][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:07:19,958][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:07:20,456][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:07:20,957][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:07:21,456][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:07:21,956][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:07:22,455][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:07:22,953][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:07:23,452][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:07:23,950][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:07:24,449][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:07:24,948][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:07:25,446][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:07:25,945][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:07:26,443][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:07:26,941][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:07:27,440][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:07:27,939][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:07:28,439][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:07:28,939][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:07:29,437][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:07:29,936][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:07:30,435][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:07:30,932][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:07:31,430][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:07:31,928][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:07:32,425][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:07:32,924][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:07:33,424][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:07:33,925][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:07:34,424][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:07:34,923][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:07:35,423][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:07:35,923][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:07:36,422][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:07:36,924][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:07:37,422][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:07:37,920][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:07:38,417][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:07:38,915][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:07:39,412][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:07:39,909][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:07:40,405][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:07:40,903][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:07:41,403][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:07:41,904][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:07:42,403][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:07:42,903][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:07:43,402][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:07:43,901][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:07:44,401][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:07:44,901][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:07:45,402][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:07:45,902][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:07:46,403][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:07:46,902][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:07:47,404][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:07:47,903][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:07:48,405][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:07:48,904][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21702 tokens. [2026-03-25 23:07:49,538][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-25 23:07:50,268][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:07:50,270][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:07:50,272][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:07:51,088][__main__][INFO] - Iteration 115 took 1m 14s (9.28% Gen, 89.62% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 31m 3s. Estimated total time: 62h 4m 12s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 8s, 500 more iterations: 10h 20m 42s. [2026-03-25 23:07:51,090][__main__][INFO] - Starting iteration 115. [2026-03-25 23:07:51,492][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:07:51,492][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:07:52,083][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:07:58,636][__main__][INFO] - Number of regex retries in iteration 115: 1 [2026-03-25 23:07:58,637][__main__][INFO] - agents played in iteration 115 are Alice, Bob [2026-03-25 23:07:59,519][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:07:59,563][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:07:59,564][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:07:59,564][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:08:00,171][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:08:00,626][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:08:01,129][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:08:01,628][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:08:02,126][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:08:02,624][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:08:03,122][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:08:03,620][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:08:04,117][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:08:04,620][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:08:05,119][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:08:05,618][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:08:06,117][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:08:06,616][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:08:07,116][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:08:07,616][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:08:08,115][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:08:08,613][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:08:09,111][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:08:09,610][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:08:10,108][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:08:10,606][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:08:11,104][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:08:11,602][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:08:12,100][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:08:12,599][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:08:13,098][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:08:13,596][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:08:14,095][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:08:14,593][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:08:15,090][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:08:15,588][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:08:16,085][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:08:16,584][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:08:17,081][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:08:17,579][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:08:18,078][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:08:18,579][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:08:19,077][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:08:19,575][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:08:20,072][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:08:20,569][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:08:21,066][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:08:21,563][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:08:22,061][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:08:22,559][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:08:23,058][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:08:23,556][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:08:24,055][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:08:24,555][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:08:25,056][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:08:25,556][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:08:26,056][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:08:26,555][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:08:27,054][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:08:27,553][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:08:28,054][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:08:28,551][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:08:29,049][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:08:29,548][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:08:30,046][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:08:30,544][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:08:31,041][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:08:31,538][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:08:32,037][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:08:32,535][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:08:33,035][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:08:33,534][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:08:34,034][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:08:34,534][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:08:35,035][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:08:35,533][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:08:36,031][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:08:36,530][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:08:37,027][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:08:37,526][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:08:38,025][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:08:38,525][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:08:39,024][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:08:39,524][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:08:40,023][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:08:40,518][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:08:41,017][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:08:41,515][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:08:42,010][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:08:42,508][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:08:43,006][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:08:43,502][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:08:44,001][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:08:44,499][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:08:44,996][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:08:45,496][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:08:45,995][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:08:46,494][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:08:46,992][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:08:47,490][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:08:47,988][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:08:48,488][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:08:48,986][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:08:49,482][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:08:49,981][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:08:50,478][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:08:50,977][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:08:51,476][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:08:51,974][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:08:52,471][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:08:52,968][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:08:53,466][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:08:53,963][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:08:54,461][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:08:54,958][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:08:55,457][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:08:55,955][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:08:56,454][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:08:56,954][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:08:57,452][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:08:57,951][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:08:58,449][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:08:58,949][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:08:59,447][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:08:59,947][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:09:00,445][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:09:00,943][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:09:01,441][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:09:01,939][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:09:02,438][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:09:02,937][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:09:03,435][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:09:03,934][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21646 tokens. [2026-03-25 23:09:04,601][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 23:09:05,335][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:09:05,337][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:09:05,339][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:09:05,991][__main__][INFO] - Iteration 116 took 1m 14s (9.59% Gen, 89.53% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 30m 35s. Estimated total time: 62h 4m 59s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 49s. [2026-03-25 23:09:05,993][__main__][INFO] - Starting iteration 116. [2026-03-25 23:09:06,394][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:09:06,395][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:09:13,085][__main__][INFO] - Number of regex retries in iteration 116: 0 [2026-03-25 23:09:13,086][__main__][INFO] - agents played in iteration 116 are Alice, Bob [2026-03-25 23:09:13,965][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:09:14,009][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:09:14,010][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:09:14,010][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:09:14,616][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:09:15,071][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:09:15,573][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:09:16,071][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:09:16,570][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:09:17,068][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:09:17,566][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:09:18,065][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:09:18,565][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:09:19,064][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:09:19,563][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:09:20,062][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:09:20,560][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:09:21,058][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:09:21,556][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:09:22,056][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:09:22,554][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:09:23,056][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:09:23,556][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:09:24,056][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:09:24,555][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:09:25,054][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:09:25,556][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:09:26,056][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:09:26,556][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:09:27,055][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:09:27,549][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:09:28,048][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:09:28,546][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:09:29,043][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:09:29,542][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:09:30,041][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:09:30,539][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:09:31,037][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:09:31,535][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:09:32,032][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:09:32,529][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:09:33,026][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:09:33,523][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:09:34,021][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:09:34,520][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:09:35,020][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:09:35,519][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:09:36,019][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:09:36,519][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:09:37,018][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:09:37,518][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:09:38,018][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:09:38,518][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:09:39,016][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:09:39,514][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:09:40,012][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:09:40,509][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:09:41,006][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:09:41,503][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:09:42,001][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:09:42,499][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:09:42,994][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:09:43,489][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:09:43,985][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:09:44,481][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:09:44,976][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:09:45,471][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:09:45,967][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:09:46,463][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:09:46,962][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:09:47,460][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:09:47,958][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:09:48,459][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:09:48,958][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:09:49,456][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:09:49,954][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:09:50,453][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:09:50,951][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:09:51,448][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:09:51,947][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:09:52,446][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:09:52,945][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:09:53,443][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:09:53,941][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:09:54,440][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:09:54,939][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:09:55,438][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:09:55,937][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:09:56,436][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:09:56,935][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:09:57,433][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:09:57,931][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:09:58,429][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:09:58,926][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:09:59,424][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:09:59,923][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:10:00,423][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:10:00,920][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:10:01,418][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:10:01,915][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:10:02,412][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:10:02,908][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:10:03,405][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:10:03,902][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:10:04,399][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:10:04,897][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:10:05,395][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:10:05,893][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:10:06,391][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:10:06,889][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:10:07,387][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:10:07,884][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:10:08,382][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:10:08,880][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:10:09,378][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:10:09,877][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:10:10,376][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:10:10,875][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:10:11,373][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:10:11,869][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:10:12,367][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:10:12,865][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:10:13,363][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:10:13,861][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:10:14,358][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:10:14,852][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:10:15,349][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:10:15,847][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:10:16,343][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:10:16,839][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:10:17,333][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:10:17,828][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:10:18,325][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21501 tokens. [2026-03-25 23:10:18,964][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.13%, Current % of VRAM taken: 60.49%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 23:10:19,706][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:10:19,709][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:10:19,710][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:10:20,517][__main__][INFO] - Iteration 117 took 1m 14s (9.03% Gen, 89.88% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 10m 31s. Estimated total time: 61h 46m 9s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 32s, 500 more iterations: 10h 17m 41s. [2026-03-25 23:10:20,519][__main__][INFO] - Starting iteration 117. [2026-03-25 23:10:20,919][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:10:20,919][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:10:27,282][__main__][INFO] - Number of regex retries in iteration 117: 0 [2026-03-25 23:10:27,283][__main__][INFO] - agents played in iteration 117 are Alice, Bob [2026-03-25 23:10:28,157][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:10:28,201][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:10:28,201][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:10:28,202][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:10:28,810][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:10:29,263][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:10:29,766][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:10:30,265][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:10:30,764][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:10:31,263][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:10:31,761][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:10:32,260][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:10:32,758][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:10:33,257][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:10:33,757][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:10:34,255][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:10:34,752][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:10:35,250][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:10:35,748][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:10:36,245][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:10:36,743][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:10:37,244][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:10:37,745][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:10:38,244][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:10:38,743][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:10:39,243][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:10:39,742][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:10:40,243][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:10:40,743][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:10:41,242][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:10:41,741][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:10:42,239][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:10:42,737][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:10:43,234][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:10:43,730][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:10:44,227][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:10:44,725][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:10:45,223][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:10:45,721][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:10:46,219][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:10:46,716][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:10:47,213][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:10:47,710][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:10:48,208][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:10:48,706][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:10:49,204][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:10:49,702][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:10:50,200][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:10:50,697][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:10:51,195][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:10:51,693][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:10:52,192][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:10:52,689][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:10:53,187][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:10:53,685][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:10:54,185][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:10:54,684][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:10:55,183][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:10:55,682][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:10:56,180][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:10:56,679][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:10:57,178][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:10:57,677][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:10:58,177][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:10:58,675][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:10:59,173][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:10:59,670][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:11:00,168][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:11:00,666][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:11:01,163][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:11:01,662][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:11:02,160][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:11:02,658][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:11:03,156][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:11:03,654][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:11:04,151][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:11:04,649][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:11:05,147][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:11:05,644][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:11:06,143][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:11:06,641][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:11:07,139][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:11:07,637][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:11:08,136][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:11:08,633][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:11:09,132][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:11:09,631][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:11:10,130][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:11:10,630][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:11:11,128][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:11:11,628][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:11:12,127][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:11:12,626][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:11:13,123][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:11:13,621][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:11:14,120][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:11:14,618][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:11:15,116][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:11:15,614][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:11:16,111][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:11:16,609][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:11:17,106][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:11:17,603][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:11:18,101][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:11:18,600][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:11:19,098][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:11:19,597][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:11:20,094][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:11:20,592][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:11:21,089][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:11:21,587][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:11:22,085][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:11:22,582][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:11:23,080][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:11:23,577][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:11:24,076][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:11:24,574][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:11:25,073][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:11:25,572][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:11:26,071][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:11:26,569][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:11:27,067][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:11:27,566][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:11:28,066][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:11:28,565][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:11:29,063][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:11:29,561][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:11:30,059][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:11:30,556][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:11:31,054][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:11:31,551][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:11:32,050][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:11:32,549][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21738 tokens. [2026-03-25 23:11:33,188][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:11:33,941][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:11:33,943][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:11:33,945][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:11:34,598][__main__][INFO] - Iteration 118 took 1m 13s (8.64% Gen, 90.47% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 47m 9s. Estimated total time: 61h 24m 1s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 48s, 500 more iterations: 10h 14m 0s. [2026-03-25 23:11:34,601][__main__][INFO] - Starting iteration 118. [2026-03-25 23:11:35,004][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:11:35,004][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:11:41,513][__main__][INFO] - Number of regex retries in iteration 118: 0 [2026-03-25 23:11:41,799][__main__][INFO] - agents played in iteration 118 are Alice, Bob [2026-03-25 23:11:42,678][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:11:42,723][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:11:42,723][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:11:42,724][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:11:43,336][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:11:43,793][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:11:44,296][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:11:44,796][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:11:45,296][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:11:45,800][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:11:46,300][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:11:46,800][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:11:47,299][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:11:47,798][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:11:48,297][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:11:48,797][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:11:49,296][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:11:49,793][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:11:50,292][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:11:50,789][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:11:51,286][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:11:51,786][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:11:52,285][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:11:52,785][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:11:53,283][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:11:53,782][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:11:54,280][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:11:54,779][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:11:55,278][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:11:55,777][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:11:56,277][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:11:56,775][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:11:57,274][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:11:57,772][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:11:58,270][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:11:58,768][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:11:59,266][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:11:59,764][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:12:00,264][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:12:00,763][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:12:01,262][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:12:01,760][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:12:02,258][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:12:02,757][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:12:03,255][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:12:03,753][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:12:04,252][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:12:04,749][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:12:05,246][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:12:05,743][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:12:06,241][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:12:06,739][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:12:07,238][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:12:07,739][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:12:08,240][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:12:08,740][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:12:09,240][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:12:09,740][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:12:10,240][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:12:10,741][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:12:11,240][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:12:11,740][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:12:12,239][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:12:12,737][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:12:13,237][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:12:13,734][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:12:14,231][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:12:14,728][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:12:15,227][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:12:15,726][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:12:16,226][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:12:16,724][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:12:17,223][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:12:17,721][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:12:18,222][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:12:18,721][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:12:19,221][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:12:19,720][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:12:20,218][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:12:20,716][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:12:21,213][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:12:21,709][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:12:22,206][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:12:22,703][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:12:23,201][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:12:23,701][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:12:24,199][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:12:24,698][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:12:25,197][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:12:25,695][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:12:26,192][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:12:26,691][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:12:27,189][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:12:27,687][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:12:28,185][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:12:28,683][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:12:29,182][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:12:29,680][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:12:30,177][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:12:30,674][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:12:31,171][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:12:31,669][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:12:32,166][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:12:32,665][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:12:33,163][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:12:33,663][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:12:34,162][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:12:34,660][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:12:35,158][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:12:35,655][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:12:36,151][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:12:36,647][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:12:37,145][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:12:37,642][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:12:38,140][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:12:38,638][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:12:39,137][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:12:39,635][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:12:40,135][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:12:40,634][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:12:41,133][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:12:41,632][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:12:42,131][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:12:42,630][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:12:43,129][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:12:43,627][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:12:44,126][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:12:44,624][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:12:45,122][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:12:45,621][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:12:46,119][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:12:46,617][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:12:47,116][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21711 tokens. [2026-03-25 23:12:47,764][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:04 [2026-03-25 23:12:48,551][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:12:48,554][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:12:48,555][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:12:49,241][__main__][INFO] - Iteration 119 took 1m 14s (9.15% Gen, 89.92% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 13m 45s. Estimated total time: 61h 51m 53s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 43s, 500 more iterations: 10h 18m 38s. [2026-03-25 23:12:49,243][__main__][INFO] - Starting iteration 119. [2026-03-25 23:12:49,644][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:12:49,645][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:12:56,284][__main__][INFO] - Number of regex retries in iteration 119: 0 [2026-03-25 23:12:56,285][__main__][INFO] - agents played in iteration 119 are Alice, Bob [2026-03-25 23:12:57,160][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:12:57,205][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:12:57,206][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:12:57,206][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:12:57,808][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:12:58,263][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:12:58,766][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:12:59,265][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:12:59,764][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:13:00,262][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:13:00,761][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:13:01,260][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:13:01,758][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:13:02,254][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:13:02,751][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:13:03,247][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:13:03,745][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:13:04,243][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:13:04,741][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:13:05,238][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:13:05,736][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:13:06,235][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:13:06,729][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:13:07,223][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:13:07,721][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:13:08,215][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:13:08,712][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:13:09,210][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:13:09,707][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:13:10,204][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:13:10,700][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:13:11,195][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:13:11,692][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:13:12,188][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:13:12,684][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:13:13,180][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:13:13,675][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:13:14,177][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:13:14,677][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:13:15,177][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:13:15,680][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:13:16,180][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:13:16,680][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:13:17,179][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:13:17,679][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:13:18,178][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:13:18,676][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:13:19,173][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:13:19,670][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:13:20,167][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:13:20,664][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:13:21,161][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:13:21,657][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:13:22,157][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:13:22,656][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:13:23,154][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:13:23,653][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:13:24,151][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:13:24,649][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:13:25,147][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:13:25,645][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:13:26,144][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:13:26,642][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:13:27,140][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:13:27,638][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:13:28,136][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:13:28,634][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:13:29,130][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:13:29,627][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:13:30,126][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:13:30,626][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:13:31,126][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:13:31,625][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:13:32,123][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:13:32,621][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:13:33,120][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:13:33,619][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:13:34,117][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:13:34,614][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:13:35,112][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:13:35,609][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:13:36,106][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:13:36,604][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:13:37,102][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:13:37,599][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:13:38,096][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:13:38,593][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:13:39,090][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:13:39,587][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:13:40,085][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:13:40,583][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:13:41,082][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:13:41,580][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:13:42,077][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:13:42,574][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:13:43,071][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:13:43,569][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:13:44,065][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:13:44,563][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:13:45,061][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:13:45,560][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:13:46,059][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:13:46,558][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:13:47,057][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:13:47,556][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:13:48,057][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:13:48,559][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:13:49,058][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:13:49,558][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:13:50,055][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:13:50,552][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:13:51,048][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:13:51,545][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:13:52,043][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:13:52,541][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:13:53,038][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:13:53,535][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:13:54,033][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:13:54,531][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:13:55,030][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:13:55,527][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:13:56,022][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:13:56,520][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:13:57,018][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:13:57,517][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:13:58,015][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:13:58,514][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:13:59,011][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:13:59,509][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:14:00,006][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:14:00,504][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:14:01,002][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:14:01,500][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21691 tokens. [2026-03-25 23:14:02,140][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 23:14:02,878][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:14:02,880][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:14:02,881][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:14:03,528][__main__][INFO] - Iteration 120 took 1m 13s (8.99% Gen, 90.14% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 54m 52s. Estimated total time: 61h 34m 13s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 8s, 500 more iterations: 10h 15m 42s. [2026-03-25 23:14:03,531][__main__][INFO] - Starting iteration 120. [2026-03-25 23:14:03,931][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:14:03,932][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:14:10,811][__main__][INFO] - Number of regex retries in iteration 120: 0 [2026-03-25 23:14:10,812][__main__][INFO] - agents played in iteration 120 are Alice, Bob [2026-03-25 23:14:11,686][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:14:11,731][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:14:11,731][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:14:11,732][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:14:12,342][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:14:12,795][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:14:13,298][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:14:13,795][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:14:14,291][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:14:14,788][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:14:15,286][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:14:15,783][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:14:16,281][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:14:16,782][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:14:17,279][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:14:17,778][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:14:18,279][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:14:18,780][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:14:19,279][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:14:19,778][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:14:20,278][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:14:20,777][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:14:21,275][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:14:21,773][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:14:22,270][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:14:22,768][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:14:23,265][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:14:23,764][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:14:24,262][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:14:24,762][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:14:25,261][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:14:25,759][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:14:26,258][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:14:26,756][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:14:27,256][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:14:27,757][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:14:28,255][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:14:28,754][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:14:29,253][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:14:29,752][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:14:30,250][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:14:30,748][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:14:31,247][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:14:31,747][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:14:32,246][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:14:32,748][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:14:33,248][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:14:33,746][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:14:34,247][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:14:34,745][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:14:35,245][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:14:35,744][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:14:36,243][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:14:36,741][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:14:37,239][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:14:37,737][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:14:38,233][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:14:38,731][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:14:39,228][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:14:39,726][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:14:40,223][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:14:40,722][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:14:41,220][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:14:41,718][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:14:42,216][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:14:42,715][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:14:43,214][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:14:43,712][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:14:44,209][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:14:44,707][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:14:45,205][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:14:45,706][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:14:46,204][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:14:46,702][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:14:47,200][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:14:47,699][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:14:48,197][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:14:48,699][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:14:49,198][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:14:49,696][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:14:50,194][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:14:50,692][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:14:51,191][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:14:51,689][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:14:52,189][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:14:52,687][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:14:53,186][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:14:53,684][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:14:54,181][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:14:54,679][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:14:55,178][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:14:55,677][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:14:56,177][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:14:56,675][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:14:57,175][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:14:57,673][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:14:58,171][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:14:58,669][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:14:59,167][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:14:59,665][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:15:00,164][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:15:00,663][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:15:01,162][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:15:01,661][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:15:02,159][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:15:02,657][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:15:03,156][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:15:03,655][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:15:04,154][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:15:04,652][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:15:05,150][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:15:05,649][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:15:06,147][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:15:06,645][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:15:07,142][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:15:07,642][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:15:08,142][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:15:08,640][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:15:09,137][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:15:09,635][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:15:10,132][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:15:10,629][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:15:11,126][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:15:11,624][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:15:12,121][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:15:12,620][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:15:13,119][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:15:13,618][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:15:14,116][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:15:14,614][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:15:15,112][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:15:15,609][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:15:16,107][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21616 tokens. [2026-03-25 23:15:16,741][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 23:15:17,476][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:15:17,479][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:15:17,480][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:15:18,125][__main__][INFO] - Iteration 121 took 1m 14s (9.27% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 9m 6s. Estimated total time: 61h 49m 42s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 17s. [2026-03-25 23:15:18,127][__main__][INFO] - Starting iteration 121. [2026-03-25 23:15:18,528][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:15:18,529][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:15:20,097][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:15:25,223][__main__][INFO] - Number of regex retries in iteration 121: 1 [2026-03-25 23:15:25,224][__main__][INFO] - agents played in iteration 121 are Alice, Bob [2026-03-25 23:15:26,098][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:15:26,142][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:15:26,143][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:15:26,143][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:15:26,749][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:15:27,205][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:15:27,708][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:15:28,207][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:15:28,705][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:15:29,204][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:15:29,704][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:15:30,203][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:15:30,702][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:15:31,203][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:15:31,703][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:15:32,202][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:15:32,701][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:15:33,202][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:15:33,701][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:15:34,201][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:15:34,700][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:15:35,198][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:15:35,696][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:15:36,193][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:15:36,691][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:15:37,188][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:15:37,686][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:15:38,184][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:15:38,681][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:15:39,180][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:15:39,678][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:15:40,177][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:15:40,674][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:15:41,173][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:15:41,671][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:15:42,170][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:15:42,669][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:15:43,167][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:15:43,666][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:15:44,165][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:15:44,664][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:15:45,162][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:15:45,661][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:15:46,160][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:15:46,658][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:15:47,156][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:15:47,654][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:15:48,151][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:15:48,651][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:15:49,148][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:15:49,646][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:15:50,143][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:15:50,641][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:15:51,139][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:15:51,636][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:15:52,133][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:15:52,631][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:15:53,129][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:15:53,626][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:15:54,124][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:15:54,622][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:15:55,121][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:15:55,619][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:15:56,119][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:15:56,618][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:15:57,117][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:15:57,614][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:15:58,113][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:15:58,611][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:15:59,110][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:15:59,609][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:16:00,108][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:16:00,607][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:16:01,106][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:16:01,604][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:16:02,103][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:16:02,602][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:16:03,101][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:16:03,601][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:16:04,100][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:16:04,600][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:16:05,101][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:16:05,600][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:16:06,100][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:16:06,600][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:16:07,099][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:16:07,598][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:16:08,096][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:16:08,593][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:16:09,091][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:16:09,589][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:16:10,088][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:16:10,586][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:16:11,085][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:16:11,584][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:16:12,083][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:16:12,581][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:16:13,080][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:16:13,578][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:16:14,079][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:16:14,577][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:16:15,075][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:16:15,576][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:16:16,075][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:16:16,573][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:16:17,071][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:16:17,569][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:16:18,068][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:16:18,568][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:16:19,067][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:16:19,565][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:16:20,060][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:16:20,557][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:16:21,054][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:16:21,550][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:16:22,047][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:16:22,545][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:16:23,042][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:16:23,539][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:16:24,037][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:16:24,534][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:16:25,032][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:16:25,529][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:16:26,027][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:16:26,524][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:16:27,023][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:16:27,521][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:16:28,020][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:16:28,519][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:16:29,018][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:16:29,516][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:16:30,015][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:16:30,512][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21725 tokens. [2026-03-25 23:16:31,141][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.50%, ΔTime: 00:01:04 [2026-03-25 23:16:31,868][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:16:31,870][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:16:31,871][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:16:32,515][__main__][INFO] - Iteration 122 took 1m 13s (9.05% Gen, 90.08% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 57m 31s. Estimated total time: 61h 39m 21s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 18s, 500 more iterations: 10h 16m 33s. [2026-03-25 23:16:32,517][__main__][INFO] - Starting iteration 122. [2026-03-25 23:16:32,920][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:16:32,920][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:16:40,583][__main__][INFO] - Number of regex retries in iteration 122: 0 [2026-03-25 23:16:40,584][__main__][INFO] - agents played in iteration 122 are Alice, Bob [2026-03-25 23:16:41,463][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:16:41,507][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:16:41,508][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:16:41,508][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:16:42,117][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:16:42,572][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:16:43,077][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:16:43,576][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:16:44,075][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:16:44,573][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:16:45,070][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:16:45,569][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:16:46,068][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:16:46,566][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:16:47,064][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:16:47,561][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:16:48,059][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:16:48,557][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:16:49,055][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:16:49,552][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:16:50,050][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:16:50,553][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:16:51,055][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:16:51,557][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:16:52,058][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:16:52,559][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:16:53,061][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:16:53,563][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:16:54,065][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:16:54,564][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:16:55,063][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:16:55,561][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:16:56,060][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:16:56,559][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:16:57,058][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:16:57,557][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:16:58,056][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:16:58,554][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:16:59,052][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:16:59,550][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:17:00,048][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:17:00,546][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:17:01,045][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:17:01,544][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:17:02,042][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:17:02,540][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:17:03,038][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:17:03,535][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:17:04,033][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:17:04,529][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:17:05,027][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:17:05,525][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:17:06,022][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:17:06,520][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:17:07,017][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:17:07,514][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:17:08,010][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:17:08,507][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:17:09,005][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:17:09,501][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:17:09,999][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:17:10,499][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:17:10,999][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:17:11,497][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:17:11,996][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:17:12,493][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:17:12,991][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:17:13,490][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:17:13,988][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:17:14,486][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:17:14,984][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:17:15,483][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:17:15,982][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:17:16,479][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:17:16,976][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:17:17,474][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:17:17,972][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:17:18,471][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:17:18,969][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:17:19,467][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:17:19,964][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:17:20,462][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:17:20,959][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:17:21,456][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:17:21,953][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:17:22,453][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:17:22,953][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:17:23,455][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:17:23,956][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:17:24,457][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:17:24,958][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:17:25,457][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:17:25,958][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:17:26,457][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:17:26,956][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:17:27,454][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:17:27,952][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:17:28,450][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:17:28,948][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:17:29,445][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:17:29,943][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:17:30,441][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:17:30,940][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:17:31,440][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:17:31,937][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:17:32,435][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:17:32,932][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:17:33,429][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:17:33,927][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:17:34,424][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:17:34,921][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:17:35,421][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:17:35,919][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:17:36,416][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:17:36,913][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:17:37,409][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:17:37,907][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:17:38,404][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:17:38,901][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:17:39,398][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:17:39,896][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:17:40,393][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:17:40,890][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:17:41,388][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:17:41,885][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:17:42,383][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:17:42,880][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:17:43,378][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:17:43,875][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:17:44,373][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:17:44,870][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:17:45,369][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:17:45,868][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21742 tokens. [2026-03-25 23:17:46,497][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-25 23:17:47,235][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:17:47,237][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:17:47,239][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:17:47,883][__main__][INFO] - Iteration 123 took 1m 14s (10.22% Gen, 88.92% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 45m 6s. Estimated total time: 62h 28m 12s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 56s, 500 more iterations: 10h 24m 42s. [2026-03-25 23:17:47,885][__main__][INFO] - Starting iteration 123. [2026-03-25 23:17:48,286][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:17:48,287][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:17:48,893][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:17:50,471][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 balls, 10 hats, 10 books did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:17:51,735][mllm.models.large_language_model_local][WARNING] - Response Since Alice values books much higher than hats or balls, and I value hats much higher than books or balls, a strategic split would be to try to capture the items I value more. Proposal: 10 hats, 10 balls, 10 books did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:17:55,506][__main__][INFO] - Number of regex retries in iteration 123: 3 [2026-03-25 23:17:55,506][__main__][INFO] - agents played in iteration 123 are Alice, Bob [2026-03-25 23:17:56,379][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:17:56,423][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:17:56,423][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:17:56,424][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:17:57,039][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:17:57,494][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:17:57,998][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:17:58,498][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:17:58,997][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:17:59,499][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:18:00,000][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:18:00,501][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:18:01,001][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:18:01,500][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:18:01,998][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:18:02,496][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:18:02,993][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:18:03,492][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:18:03,990][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:18:04,489][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:18:04,987][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:18:05,486][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:18:05,984][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:18:06,483][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:18:06,982][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:18:07,481][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:18:07,980][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:18:08,478][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:18:08,977][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:18:09,472][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:18:09,969][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:18:10,465][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:18:10,961][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:18:11,457][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:18:11,954][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:18:12,450][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:18:12,948][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:18:13,448][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:18:13,947][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:18:14,446][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:18:14,945][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:18:15,444][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:18:15,942][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:18:16,442][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:18:16,941][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:18:17,440][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:18:17,940][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:18:18,441][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:18:18,940][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:18:19,441][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:18:19,941][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:18:20,441][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:18:20,940][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:18:21,440][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:18:21,937][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:18:22,435][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:18:22,932][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:18:23,429][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:18:23,926][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:18:24,425][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:18:24,922][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:18:25,421][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:18:25,919][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:18:26,417][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:18:26,914][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:18:27,412][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:18:27,908][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:18:28,406][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:18:28,904][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:18:29,404][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:18:29,903][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:18:30,403][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:18:30,902][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:18:31,401][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:18:31,900][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:18:32,401][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:18:32,901][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:18:33,399][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:18:33,898][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:18:34,396][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:18:34,893][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:18:35,390][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:18:35,888][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:18:36,386][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:18:36,884][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:18:37,382][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:18:37,880][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:18:38,379][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:18:38,877][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:18:39,374][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:18:39,871][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:18:40,369][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:18:40,867][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:18:41,363][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:18:41,859][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:18:42,355][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:18:42,851][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:18:43,347][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:18:43,843][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:18:44,339][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:18:44,835][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:18:45,334][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:18:45,832][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:18:46,331][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:18:46,831][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:18:47,330][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:18:47,830][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:18:48,330][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:18:48,829][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:18:49,328][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:18:49,826][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:18:50,326][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:18:50,825][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:18:51,324][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:18:51,823][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:18:52,322][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:18:52,821][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:18:53,319][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:18:53,817][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:18:54,315][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:18:54,813][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:18:55,311][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:18:55,808][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:18:56,306][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:18:56,803][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:18:57,301][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:18:57,799][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:18:58,298][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:18:58,796][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:18:59,293][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:18:59,791][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:19:00,288][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:19:00,785][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-25 23:19:01,422][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:19:02,156][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:19:02,158][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:19:02,159][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:19:03,025][__main__][INFO] - Iteration 124 took 1m 14s (9.66% Gen, 89.18% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 32m 36s. Estimated total time: 62h 16m 57s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 33s, 500 more iterations: 10h 22m 49s. [2026-03-25 23:19:03,027][__main__][INFO] - Starting iteration 124. [2026-03-25 23:19:03,425][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:19:03,426][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:19:04,498][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:19:10,405][__main__][INFO] - Number of regex retries in iteration 124: 1 [2026-03-25 23:19:10,406][__main__][INFO] - agents played in iteration 124 are Alice, Bob [2026-03-25 23:19:11,285][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:19:11,329][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:19:11,330][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:19:11,330][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:19:11,939][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:19:12,395][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:19:12,897][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:19:13,395][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:19:13,893][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:19:14,392][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:19:14,889][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:19:15,389][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:19:15,890][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:19:16,389][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:19:16,888][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:19:17,386][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:19:17,886][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:19:18,387][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:19:18,886][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:19:19,385][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:19:19,884][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:19:20,385][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:19:20,884][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:19:21,384][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:19:21,884][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:19:22,383][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:19:22,884][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:19:23,385][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:19:23,885][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:19:24,383][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:19:24,880][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:19:25,379][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:19:25,877][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:19:26,375][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:19:26,872][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:19:27,368][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:19:27,867][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:19:28,366][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:19:28,865][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:19:29,363][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:19:29,861][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:19:30,361][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:19:30,860][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:19:31,359][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:19:31,858][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:19:32,358][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:19:32,858][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:19:33,357][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:19:33,858][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:19:34,359][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:19:34,859][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:19:35,359][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:19:35,859][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:19:36,355][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:19:36,853][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:19:37,349][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:19:37,849][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:19:38,348][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:19:38,846][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:19:39,344][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:19:39,842][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:19:40,339][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:19:40,837][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:19:41,334][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:19:41,831][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:19:42,329][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:19:42,827][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:19:43,325][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:19:43,823][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:19:44,320][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:19:44,817][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:19:45,314][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:19:45,812][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:19:46,309][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:19:46,807][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:19:47,305][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:19:47,805][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:19:48,303][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:19:48,802][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:19:49,301][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:19:49,799][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:19:50,296][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:19:50,794][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:19:51,291][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:19:51,789][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:19:52,289][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:19:52,789][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:19:53,287][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:19:53,786][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:19:54,286][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:19:54,785][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:19:55,284][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:19:55,784][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:19:56,282][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:19:56,780][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:19:57,277][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:19:57,774][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:19:58,271][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:19:58,769][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:19:59,266][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:19:59,764][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:20:00,263][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:20:00,761][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:20:01,260][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:20:01,758][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:20:02,256][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:20:02,754][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:20:03,252][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:20:03,750][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:20:04,251][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:20:04,751][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:20:05,250][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:20:05,749][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:20:06,249][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:20:06,748][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:20:07,247][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:20:07,747][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:20:08,246][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:20:08,745][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:20:09,242][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:20:09,741][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:20:10,239][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:20:10,736][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:20:11,235][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:20:11,732][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:20:12,230][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:20:12,729][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:20:13,227][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:20:13,725][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:20:14,224][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:20:14,723][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:20:15,221][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:20:15,721][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21635 tokens. [2026-03-25 23:20:16,358][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-25 23:20:17,088][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:20:17,091][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:20:17,092][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:20:17,737][__main__][INFO] - Iteration 125 took 1m 14s (9.39% Gen, 89.74% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 59h 10m 2s. Estimated total time: 61h 55m 38s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 51s, 500 more iterations: 10h 19m 16s. [2026-03-25 23:20:17,740][__main__][INFO] - Starting iteration 125. [2026-03-25 23:20:18,144][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:20:18,145][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:20:24,760][__main__][INFO] - Number of regex retries in iteration 125: 0 [2026-03-25 23:20:24,761][__main__][INFO] - agents played in iteration 125 are Alice, Bob [2026-03-25 23:20:25,633][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:20:25,677][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:20:25,678][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:20:25,678][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:20:26,281][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:20:26,736][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:20:27,238][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:20:27,735][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:20:28,234][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:20:28,731][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:20:29,229][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:20:29,727][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:20:30,226][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:20:30,724][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:20:31,222][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:20:31,719][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:20:32,218][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:20:32,715][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:20:33,213][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:20:33,710][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:20:34,209][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:20:34,710][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:20:35,210][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:20:35,709][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:20:36,208][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:20:36,707][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:20:37,207][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:20:37,706][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:20:38,206][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:20:38,704][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:20:39,202][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:20:39,700][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:20:40,199][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:20:40,696][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:20:41,193][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:20:41,691][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:20:42,189][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:20:42,688][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:20:43,186][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:20:43,684][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:20:44,182][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:20:44,681][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:20:45,180][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:20:45,683][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:20:46,182][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:20:46,682][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:20:47,181][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:20:47,680][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:20:48,181][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:20:48,681][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:20:49,181][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:20:49,683][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:20:50,183][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:20:50,681][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:20:51,179][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:20:51,677][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:20:52,174][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:20:52,671][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:20:53,168][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:20:53,666][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:20:54,164][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:20:54,662][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:20:55,160][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:20:55,658][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:20:56,156][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:20:56,652][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:20:57,149][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:20:57,647][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:20:58,145][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:20:58,644][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:20:59,142][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:20:59,641][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:21:00,139][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:21:00,637][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:21:01,134][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:21:01,632][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:21:02,130][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:21:02,628][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:21:03,126][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:21:03,623][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:21:04,121][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:21:04,618][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:21:05,115][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:21:05,612][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:21:06,109][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:21:06,608][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:21:07,108][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:21:07,608][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:21:08,109][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:21:08,608][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:21:09,107][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:21:09,607][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:21:10,107][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:21:10,605][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:21:11,103][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:21:11,602][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:21:12,100][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:21:12,598][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:21:13,096][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:21:13,594][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:21:14,092][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:21:14,589][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:21:15,088][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:21:15,587][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:21:16,087][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:21:16,584][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:21:17,082][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:21:17,581][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:21:18,079][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:21:18,581][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:21:19,082][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:21:19,582][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:21:20,081][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:21:20,580][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:21:21,079][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:21:21,578][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:21:22,077][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:21:22,575][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:21:23,072][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:21:23,569][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:21:24,066][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:21:24,563][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:21:25,061][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:21:25,559][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:21:26,055][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:21:26,553][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:21:27,050][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:21:27,548][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:21:28,045][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:21:28,543][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:21:29,041][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:21:29,538][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:21:30,035][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-25 23:21:30,673][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 23:21:31,405][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:21:31,407][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:21:31,408][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:21:32,063][__main__][INFO] - Iteration 126 took 1m 13s (8.95% Gen, 90.16% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 49m 7s. Estimated total time: 61h 35m 57s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 11s, 500 more iterations: 10h 15m 59s. [2026-03-25 23:21:32,065][__main__][INFO] - Starting iteration 126. [2026-03-25 23:21:32,466][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:21:32,466][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:21:38,680][__main__][INFO] - Number of regex retries in iteration 126: 0 [2026-03-25 23:21:38,681][__main__][INFO] - agents played in iteration 126 are Alice, Bob [2026-03-25 23:21:39,555][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:21:39,600][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:21:39,600][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:21:39,601][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:21:40,210][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:21:40,662][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:21:41,165][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:21:41,661][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:21:42,157][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:21:42,653][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:21:43,150][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:21:43,647][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:21:44,144][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:21:44,645][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:21:45,144][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:21:45,643][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:21:46,143][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:21:46,642][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:21:47,143][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:21:47,642][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:21:48,141][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:21:48,642][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:21:49,140][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:21:49,640][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:21:50,138][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:21:50,637][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:21:51,136][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:21:51,635][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:21:52,133][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:21:52,631][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:21:53,129][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:21:53,628][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:21:54,126][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:21:54,623][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:21:55,124][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:21:55,623][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:21:56,122][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:21:56,619][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:21:57,118][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:21:57,616][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:21:58,112][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:21:58,609][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:21:59,106][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:21:59,604][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:22:00,102][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:22:00,601][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:22:01,100][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:22:01,597][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:22:02,096][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:22:02,593][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:22:03,090][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:22:03,589][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:22:04,087][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:22:04,586][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:22:05,084][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:22:05,582][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:22:06,081][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:22:06,579][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:22:07,077][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:22:07,576][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:22:08,073][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:22:08,570][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:22:09,069][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:22:09,568][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:22:10,067][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:22:10,566][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:22:11,065][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:22:11,563][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:22:12,063][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:22:12,559][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:22:13,055][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:22:13,553][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:22:14,049][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:22:14,546][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:22:15,042][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:22:15,541][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:22:16,038][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:22:16,538][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:22:17,035][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:22:17,535][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:22:18,034][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:22:18,534][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:22:19,033][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:22:19,530][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:22:20,029][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:22:20,528][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:22:21,026][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:22:21,526][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:22:22,025][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:22:22,523][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:22:23,022][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:22:23,521][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:22:24,020][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:22:24,519][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:22:25,018][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:22:25,516][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:22:26,014][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:22:26,512][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:22:27,010][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:22:27,509][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:22:28,007][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:22:28,507][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:22:29,006][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:22:29,504][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:22:30,002][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:22:30,500][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:22:30,997][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:22:31,495][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:22:31,992][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:22:32,490][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:22:32,989][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:22:33,487][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:22:33,985][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:22:34,484][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:22:34,982][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:22:35,481][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:22:35,979][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:22:36,476][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:22:36,974][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:22:37,473][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:22:37,971][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:22:38,469][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:22:38,968][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:22:39,468][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:22:39,966][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:22:40,464][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:22:40,962][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:22:41,462][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:22:41,961][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:22:42,459][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:22:42,958][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:22:43,457][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:22:43,956][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-25 23:22:44,585][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-25 23:22:45,315][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:22:45,318][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:22:45,319][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:22:45,966][__main__][INFO] - Iteration 127 took 1m 13s (8.45% Gen, 90.66% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 26m 59s. Estimated total time: 61h 15m 3s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 30s, 500 more iterations: 10h 12m 30s. [2026-03-25 23:22:45,968][__main__][INFO] - Starting iteration 127. [2026-03-25 23:22:46,368][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:22:46,369][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:22:53,092][__main__][INFO] - Number of regex retries in iteration 127: 0 [2026-03-25 23:22:53,093][__main__][INFO] - agents played in iteration 127 are Alice, Bob [2026-03-25 23:22:53,966][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:22:54,010][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:22:54,010][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:22:54,011][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:22:54,614][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:22:55,068][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:22:55,572][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:22:56,070][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:22:56,569][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:22:57,067][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:22:57,567][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:22:58,067][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:22:58,566][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:22:59,063][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:22:59,560][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:23:00,058][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:23:00,554][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:23:01,051][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:23:01,549][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:23:02,047][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:23:02,544][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:23:03,043][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:23:03,542][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:23:04,039][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:23:04,537][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:23:05,035][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:23:05,532][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:23:06,031][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:23:06,530][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:23:07,028][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:23:07,525][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:23:08,022][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:23:08,519][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:23:09,017][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:23:09,514][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:23:10,010][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:23:10,505][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:23:11,003][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:23:11,501][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:23:11,998][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:23:12,495][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:23:12,992][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:23:13,489][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:23:13,986][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:23:14,482][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:23:14,982][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:23:15,481][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:23:15,980][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:23:16,479][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:23:16,976][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:23:17,475][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:23:17,973][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:23:18,474][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:23:18,972][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:23:19,468][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:23:19,966][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:23:20,463][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:23:20,960][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:23:21,458][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:23:21,955][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:23:22,453][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:23:22,951][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:23:23,449][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:23:23,947][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:23:24,446][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:23:24,944][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:23:25,443][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:23:25,942][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:23:26,439][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:23:26,937][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:23:27,435][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:23:27,932][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:23:28,431][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:23:28,930][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:23:29,428][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:23:29,926][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:23:30,425][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:23:30,922][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:23:31,419][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:23:31,915][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:23:32,409][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:23:32,905][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:23:33,402][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:23:33,900][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:23:34,397][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:23:34,895][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:23:35,392][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:23:35,890][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:23:36,388][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:23:36,886][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:23:37,383][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:23:37,882][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:23:38,380][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:23:38,878][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:23:39,376][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:23:39,873][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:23:40,371][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:23:40,868][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:23:41,365][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:23:41,862][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:23:42,359][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:23:42,857][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:23:43,353][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:23:43,851][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:23:44,348][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:23:44,845][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:23:45,344][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:23:45,846][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:23:46,344][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:23:46,844][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:23:47,343][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:23:47,842][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:23:48,342][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:23:48,841][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:23:49,339][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:23:49,839][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:23:50,337][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:23:50,833][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:23:51,331][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:23:51,829][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:23:52,327][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:23:52,825][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:23:53,323][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:23:53,820][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:23:54,316][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:23:54,813][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:23:55,311][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:23:55,809][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:23:56,307][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:23:56,805][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:23:57,302][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:23:57,800][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:23:58,298][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21707 tokens. [2026-03-25 23:23:58,928][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 23:23:59,657][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:23:59,659][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:23:59,661][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:24:00,309][__main__][INFO] - Iteration 128 took 1m 13s (9.09% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 47m 45s. Estimated total time: 61h 37m 3s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 14s, 500 more iterations: 10h 16m 10s. [2026-03-25 23:24:00,311][__main__][INFO] - Starting iteration 128. [2026-03-25 23:24:00,714][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:24:00,715][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:24:07,275][__main__][INFO] - Number of regex retries in iteration 128: 0 [2026-03-25 23:24:07,276][__main__][INFO] - agents played in iteration 128 are Alice, Bob [2026-03-25 23:24:08,149][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:24:08,193][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:24:08,194][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:24:08,194][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:24:08,801][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:24:09,257][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:24:09,760][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:24:10,260][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:24:10,758][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:24:11,256][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:24:11,754][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:24:12,253][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:24:12,751][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:24:13,250][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:24:13,748][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:24:14,245][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:24:14,744][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:24:15,242][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:24:15,740][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:24:16,238][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:24:16,736][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:24:17,234][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:24:17,733][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:24:18,232][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:24:18,731][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:24:19,229][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:24:19,728][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:24:20,226][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:24:20,725][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:24:21,223][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:24:21,722][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:24:22,221][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:24:22,719][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:24:23,217][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:24:23,714][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:24:24,211][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:24:24,709][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:24:25,208][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:24:25,706][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:24:26,203][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:24:26,702][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:24:27,199][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:24:27,698][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:24:28,195][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:24:28,693][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:24:29,190][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:24:29,688][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:24:30,185][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:24:30,684][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:24:31,183][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:24:31,680][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:24:32,177][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:24:32,674][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:24:33,172][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:24:33,671][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:24:34,169][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:24:34,668][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:24:35,167][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:24:35,665][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:24:36,162][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:24:36,660][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:24:37,157][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:24:37,655][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:24:38,154][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:24:38,651][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:24:39,151][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:24:39,649][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:24:40,147][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:24:40,646][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:24:41,146][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:24:41,645][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:24:42,143][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:24:42,643][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:24:43,141][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:24:43,640][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:24:44,137][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:24:44,635][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:24:45,133][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:24:45,630][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:24:46,128][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:24:46,626][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:24:47,124][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:24:47,622][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:24:48,120][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:24:48,617][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:24:49,115][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:24:49,613][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:24:50,110][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:24:50,608][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:24:51,107][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:24:51,605][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:24:52,103][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:24:52,602][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:24:53,100][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:24:53,600][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:24:54,098][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:24:54,596][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:24:55,095][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:24:55,594][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:24:56,092][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:24:56,590][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:24:57,088][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:24:57,585][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:24:58,083][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:24:58,581][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:24:59,079][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:24:59,576][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:25:00,073][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:25:00,571][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:25:01,069][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:25:01,567][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:25:02,065][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:25:02,562][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:25:03,059][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:25:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:25:04,055][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:25:04,552][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:25:05,052][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:25:05,549][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:25:06,047][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:25:06,547][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:25:07,046][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:25:07,544][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:25:08,042][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:25:08,541][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:25:09,041][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:25:09,539][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:25:10,037][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:25:10,534][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:25:11,032][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:25:11,531][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:25:12,029][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:25:12,528][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-25 23:25:13,162][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 23:25:13,894][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:25:13,896][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:25:13,898][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:25:14,718][__main__][INFO] - Iteration 129 took 1m 14s (8.87% Gen, 90.02% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 49m 41s. Estimated total time: 61h 40m 14s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 20s, 500 more iterations: 10h 16m 42s. [2026-03-25 23:25:14,721][__main__][INFO] - Starting iteration 129. [2026-03-25 23:25:15,122][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:25:15,122][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:25:21,651][__main__][INFO] - Number of regex retries in iteration 129: 0 [2026-03-25 23:25:21,652][__main__][INFO] - agents played in iteration 129 are Alice, Bob [2026-03-25 23:25:22,525][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:25:22,569][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:25:22,570][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:25:22,570][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:25:23,180][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:25:23,636][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:25:24,140][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:25:24,639][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:25:25,138][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:25:25,637][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:25:26,135][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:25:26,634][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:25:27,133][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:25:27,631][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:25:28,129][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:25:28,628][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:25:29,127][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:25:29,625][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:25:30,123][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:25:30,620][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:25:31,119][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:25:31,617][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:25:32,116][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:25:32,614][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:25:33,112][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:25:33,610][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:25:34,108][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:25:34,606][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:25:35,105][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:25:35,604][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:25:36,101][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:25:36,600][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:25:37,100][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:25:37,599][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:25:38,096][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:25:38,595][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:25:39,093][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:25:39,591][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:25:40,089][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:25:40,588][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:25:41,086][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:25:41,584][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:25:42,082][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:25:42,582][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:25:43,080][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:25:43,577][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:25:44,074][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:25:44,573][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:25:45,072][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:25:45,570][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:25:46,068][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:25:46,567][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:25:47,066][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:25:47,567][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:25:48,067][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:25:48,567][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:25:49,067][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:25:49,566][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:25:50,066][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:25:50,566][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:25:51,065][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:25:51,562][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:25:52,060][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:25:52,559][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:25:53,057][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:25:53,556][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:25:54,054][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:25:54,552][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:25:55,053][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:25:55,552][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:25:56,051][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:25:56,549][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:25:57,049][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:25:57,549][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:25:58,048][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:25:58,546][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:25:59,045][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:25:59,544][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:26:00,042][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:26:00,542][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:26:01,042][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:26:01,541][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:26:02,039][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:26:02,536][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:26:03,034][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:26:03,531][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:26:04,029][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:26:04,526][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:26:05,025][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:26:05,523][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:26:06,021][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:26:06,519][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:26:07,016][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:26:07,514][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:26:08,011][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:26:08,509][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:26:09,007][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:26:09,506][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:26:10,004][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:26:10,502][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:26:11,000][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:26:11,496][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:26:11,993][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:26:12,491][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:26:12,989][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:26:13,487][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:26:13,985][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:26:14,483][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:26:14,982][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:26:15,481][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:26:15,978][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:26:16,476][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:26:16,973][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:26:17,472][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:26:17,971][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:26:18,471][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:26:18,969][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:26:19,469][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:26:19,968][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:26:20,466][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:26:20,966][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:26:21,465][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:26:21,964][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:26:22,463][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:26:22,964][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:26:23,463][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:26:23,960][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:26:24,457][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:26:24,954][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:26:25,452][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:26:25,949][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:26:26,448][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:26:26,945][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21750 tokens. [2026-03-25 23:26:27,582][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-25 23:26:28,318][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:26:28,320][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:26:28,322][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:26:28,976][__main__][INFO] - Iteration 130 took 1m 13s (8.84% Gen, 90.27% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 40m 58s. Estimated total time: 61h 32m 45s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 5s, 500 more iterations: 10h 15m 27s. [2026-03-25 23:26:28,978][__main__][INFO] - Starting iteration 130. [2026-03-25 23:26:29,378][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:26:29,379][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:26:29,968][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:26:35,985][__main__][INFO] - Number of regex retries in iteration 130: 1 [2026-03-25 23:26:35,986][__main__][INFO] - agents played in iteration 130 are Alice, Bob [2026-03-25 23:26:36,861][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:26:36,905][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:26:36,906][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:26:36,906][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:26:37,524][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:26:37,980][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:26:38,484][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:26:38,985][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:26:39,485][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:26:39,985][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:26:40,484][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:26:40,984][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:26:41,484][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:26:41,985][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:26:42,485][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:26:42,984][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:26:43,485][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:26:43,984][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:26:44,483][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:26:44,981][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:26:45,484][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:26:45,987][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:26:46,487][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:26:46,987][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:26:47,486][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:26:47,987][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:26:48,487][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:26:48,987][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:26:49,487][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:26:49,988][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:26:50,487][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:26:50,986][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:26:51,486][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:26:51,986][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:26:52,487][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:26:52,987][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:26:53,489][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:26:53,988][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:26:54,486][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:26:54,986][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:26:55,485][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:26:55,985][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:26:56,484][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:26:56,984][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:26:57,483][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:26:57,981][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:26:58,480][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:26:58,978][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:26:59,477][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:26:59,974][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:27:00,473][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:27:00,972][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:27:01,470][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:27:01,968][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:27:02,466][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:27:02,963][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:27:03,460][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:27:03,957][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:27:04,457][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:27:04,954][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:27:05,453][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:27:05,952][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:27:06,451][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:27:06,949][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:27:07,448][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:27:07,947][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:27:08,447][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:27:08,946][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:27:09,446][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:27:09,945][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:27:10,445][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:27:10,946][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:27:11,445][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:27:11,945][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:27:12,445][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:27:12,945][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:27:13,445][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:27:13,943][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:27:14,443][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:27:14,945][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:27:15,444][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:27:15,944][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:27:16,444][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:27:16,943][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:27:17,441][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:27:17,941][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:27:18,442][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:27:18,942][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:27:19,443][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:27:19,943][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:27:20,443][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:27:20,941][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:27:21,440][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:27:21,939][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:27:22,438][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:27:22,937][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:27:23,436][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:27:23,936][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:27:24,435][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:27:24,935][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:27:25,435][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:27:25,934][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:27:26,433][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:27:26,932][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:27:27,432][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:27:27,931][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:27:28,429][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:27:28,928][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:27:29,427][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:27:29,927][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:27:30,426][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:27:30,926][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:27:31,426][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:27:31,924][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:27:32,424][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:27:32,923][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:27:33,423][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:27:33,922][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:27:34,420][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:27:34,919][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:27:35,416][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:27:35,913][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:27:36,411][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:27:36,909][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:27:37,406][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:27:37,907][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:27:38,406][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:27:38,905][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:27:39,405][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:27:39,905][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:27:40,403][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:27:40,902][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:27:41,401][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21745 tokens. [2026-03-25 23:27:42,046][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-25 23:27:42,790][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:27:42,793][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:27:42,794][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:27:43,502][__main__][INFO] - Iteration 131 took 1m 14s (8.91% Gen, 90.13% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 53m 12s. Estimated total time: 61h 46m 13s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 32s, 500 more iterations: 10h 17m 42s. [2026-03-25 23:27:43,504][__main__][INFO] - Starting iteration 131. [2026-03-25 23:27:43,905][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:27:43,906][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:27:48,919][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:27:49,697][__main__][INFO] - Number of regex retries in iteration 131: 1 [2026-03-25 23:27:49,985][__main__][INFO] - agents played in iteration 131 are Alice, Bob [2026-03-25 23:27:50,863][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:27:50,908][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:27:50,908][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:27:50,909][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:27:51,527][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:27:51,984][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:27:52,488][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:27:52,988][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:27:53,488][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:27:53,986][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:27:54,485][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:27:54,985][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:27:55,486][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:27:55,988][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:27:56,489][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:27:56,989][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:27:57,489][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:27:57,989][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:27:58,489][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:27:58,988][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:27:59,488][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:27:59,986][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:28:00,484][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:28:00,982][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:28:01,479][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:28:01,977][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:28:02,474][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:28:02,972][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:28:03,470][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:28:03,968][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:28:04,466][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:28:04,964][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:28:05,461][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:28:05,960][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:28:06,457][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:28:06,954][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:28:07,453][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:28:07,953][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:28:08,453][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:28:08,952][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:28:09,451][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:28:09,951][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:28:10,450][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:28:10,951][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:28:11,451][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:28:11,950][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:28:12,449][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:28:12,947][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:28:13,447][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:28:13,947][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:28:14,445][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:28:14,945][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:28:15,446][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:28:15,948][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:28:16,447][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:28:16,945][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:28:17,445][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:28:17,945][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:28:18,447][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:28:18,945][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:28:19,444][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:28:19,944][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:28:20,441][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:28:20,939][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:28:21,436][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:28:21,934][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:28:22,432][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:28:22,932][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:28:23,430][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:28:23,931][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:28:24,429][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:28:24,928][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:28:25,428][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:28:25,928][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:28:26,428][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:28:26,927][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:28:27,427][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:28:27,926][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:28:28,426][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:28:28,926][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:28:29,428][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:28:29,928][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:28:30,428][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:28:30,928][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:28:31,427][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:28:31,925][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:28:32,423][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:28:32,921][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:28:33,418][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:28:33,915][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:28:34,412][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:28:34,910][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:28:35,408][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:28:35,906][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:28:36,404][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:28:36,901][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:28:37,398][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:28:37,894][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:28:38,392][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:28:38,889][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:28:39,387][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:28:39,887][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:28:40,385][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:28:40,884][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:28:41,384][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:28:41,883][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:28:42,383][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:28:42,884][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:28:43,384][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:28:43,882][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:28:44,380][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:28:44,878][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:28:45,376][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:28:45,875][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:28:46,373][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:28:46,873][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:28:47,372][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:28:47,871][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:28:48,372][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:28:48,870][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:28:49,370][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:28:49,869][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:28:50,368][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:28:50,867][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:28:51,366][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:28:51,864][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:28:52,361][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:28:52,858][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:28:53,356][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:28:53,853][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:28:54,350][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:28:54,849][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:28:55,348][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21738 tokens. [2026-03-25 23:28:55,993][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-25 23:28:56,743][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:28:56,745][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:28:56,798][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:28:57,657][__main__][INFO] - Iteration 132 took 1m 13s (8.24% Gen, 90.59% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 33m 23s. Estimated total time: 61h 27m 39s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 55s, 500 more iterations: 10h 14m 36s. [2026-03-25 23:28:57,661][__main__][INFO] - Starting iteration 132. [2026-03-25 23:28:58,064][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:28:58,065][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:29:04,703][__main__][INFO] - Number of regex retries in iteration 132: 0 [2026-03-25 23:29:04,704][__main__][INFO] - agents played in iteration 132 are Alice, Bob [2026-03-25 23:29:05,577][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:29:05,621][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:29:05,622][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:29:05,623][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:29:06,225][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:29:06,681][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:29:07,185][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:29:07,684][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:29:08,182][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:29:08,680][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:29:09,178][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:29:09,676][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:29:10,174][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:29:10,672][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:29:11,170][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:29:11,668][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:29:12,167][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:29:12,665][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:29:13,163][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:29:13,659][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:29:14,158][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:29:14,654][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:29:15,151][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:29:15,648][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:29:16,145][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:29:16,642][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:29:17,139][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:29:17,636][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:29:18,133][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:29:18,634][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:29:19,133][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:29:19,633][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:29:20,132][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:29:20,631][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:29:21,130][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:29:21,630][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:29:22,129][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:29:22,630][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:29:23,131][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:29:23,631][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:29:24,131][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:29:24,630][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:29:25,130][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:29:25,631][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:29:26,131][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:29:26,630][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:29:27,127][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:29:27,626][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:29:28,124][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:29:28,621][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:29:29,119][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:29:29,616][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:29:30,113][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:29:30,611][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:29:31,110][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:29:31,608][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:29:32,106][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:29:32,605][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:29:33,103][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:29:33,601][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:29:34,098][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:29:34,597][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:29:35,096][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:29:35,595][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:29:36,094][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:29:36,593][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:29:37,092][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:29:37,591][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:29:38,091][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:29:38,589][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:29:39,088][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:29:39,587][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:29:40,085][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:29:40,585][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:29:41,082][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:29:41,581][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:29:42,080][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:29:42,577][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:29:43,076][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:29:43,575][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:29:44,074][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:29:44,573][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:29:45,071][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:29:45,573][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:29:46,075][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:29:46,572][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:29:47,070][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:29:47,567][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:29:48,066][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:29:48,565][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:29:49,062][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:29:49,559][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:29:50,055][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:29:50,554][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:29:51,054][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:29:51,555][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:29:52,054][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:29:52,554][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:29:53,054][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:29:53,554][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:29:54,053][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:29:54,554][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:29:55,055][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:29:55,555][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:29:56,055][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:29:56,554][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:29:57,053][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:29:57,553][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:29:58,054][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:29:58,553][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:29:59,052][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:29:59,550][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:30:00,048][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:30:00,546][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:30:01,045][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:30:01,545][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:30:02,044][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:30:02,543][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:30:03,042][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:30:03,542][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:30:04,039][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:30:04,537][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:30:05,035][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:30:05,533][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:30:06,033][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:30:06,531][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:30:07,030][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:30:07,529][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:30:08,027][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:30:08,526][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:30:09,025][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:30:09,524][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:30:10,024][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-25 23:30:10,664][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:30:11,402][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:30:11,404][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:30:11,406][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:30:12,105][__main__][INFO] - Iteration 133 took 1m 14s (8.97% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 46m 35s. Estimated total time: 61h 42m 5s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 24s, 500 more iterations: 10h 17m 0s. [2026-03-25 23:30:12,108][__main__][INFO] - Starting iteration 133. [2026-03-25 23:30:12,513][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:30:12,514][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:30:14,186][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:30:19,349][__main__][INFO] - Number of regex retries in iteration 133: 1 [2026-03-25 23:30:19,350][__main__][INFO] - agents played in iteration 133 are Alice, Bob [2026-03-25 23:30:20,230][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:30:20,274][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:30:20,274][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:30:20,275][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:30:20,881][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:30:21,338][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:30:21,840][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:30:22,339][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:30:22,836][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:30:23,334][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:30:23,833][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:30:24,331][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:30:24,830][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:30:25,329][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:30:25,827][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:30:26,324][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:30:26,824][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:30:27,323][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:30:27,821][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:30:28,318][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:30:28,815][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:30:29,314][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:30:29,812][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:30:30,310][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:30:30,808][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:30:31,307][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:30:31,805][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:30:32,303][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:30:32,802][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:30:33,299][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:30:33,797][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:30:34,295][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:30:34,792][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:30:35,290][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:30:35,787][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:30:36,285][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:30:36,784][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:30:37,285][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:30:37,785][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:30:38,283][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:30:38,781][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:30:39,279][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:30:39,778][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:30:40,277][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:30:40,776][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:30:41,273][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:30:41,771][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:30:42,269][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:30:42,767][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:30:43,265][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:30:43,764][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:30:44,263][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:30:44,760][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:30:45,259][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:30:45,756][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:30:46,254][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:30:46,752][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:30:47,250][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:30:47,748][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:30:48,248][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:30:48,747][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:30:49,245][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:30:49,744][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:30:50,242][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:30:50,740][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:30:51,237][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:30:51,734][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:30:52,233][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:30:52,731][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:30:53,230][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:30:53,728][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:30:54,226][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:30:54,724][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:30:55,222][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:30:55,720][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:30:56,217][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:30:56,715][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:30:57,213][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:30:57,710][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:30:58,208][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:30:58,705][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:30:59,201][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:30:59,699][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:31:00,196][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:31:00,694][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:31:01,192][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:31:01,690][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:31:02,188][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:31:02,686][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:31:03,185][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:31:03,684][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:31:04,182][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:31:04,680][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:31:05,177][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:31:05,674][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:31:06,171][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:31:06,668][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:31:07,166][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:31:07,663][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:31:08,160][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:31:08,658][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:31:09,156][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:31:09,654][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:31:10,153][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:31:10,653][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:31:11,154][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:31:11,654][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:31:12,153][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:31:12,651][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:31:13,150][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:31:13,648][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:31:14,146][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:31:14,645][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:31:15,143][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:31:15,646][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:31:16,145][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:31:16,644][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:31:17,142][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:31:17,639][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:31:18,137][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:31:18,636][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:31:19,134][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:31:19,633][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:31:20,131][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:31:20,630][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:31:21,129][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:31:21,626][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:31:22,124][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:31:22,622][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:31:23,119][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:31:23,617][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:31:24,113][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:31:24,611][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21727 tokens. [2026-03-25 23:31:25,244][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 23:31:25,973][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:31:25,976][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:31:25,977][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:31:26,636][__main__][INFO] - Iteration 134 took 1m 14s (9.22% Gen, 89.89% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 49m 26s. Estimated total time: 61h 46m 11s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 32s, 500 more iterations: 10h 17m 41s. [2026-03-25 23:31:26,638][__main__][INFO] - Starting iteration 134. [2026-03-25 23:31:27,040][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:31:27,041][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:31:28,240][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:31:29,936][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:31:34,349][__main__][INFO] - Number of regex retries in iteration 134: 2 [2026-03-25 23:31:34,350][__main__][INFO] - agents played in iteration 134 are Alice, Bob [2026-03-25 23:31:35,221][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:31:35,265][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:31:35,266][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:31:35,267][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:31:35,879][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:31:36,337][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:31:36,841][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:31:37,338][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:31:37,835][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:31:38,332][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:31:38,831][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:31:39,329][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:31:39,826][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:31:40,326][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:31:40,824][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:31:41,322][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:31:41,820][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:31:42,317][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:31:42,814][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:31:43,312][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:31:43,809][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:31:44,307][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:31:44,806][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:31:45,304][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:31:45,802][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:31:46,299][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:31:46,796][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:31:47,293][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:31:47,791][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:31:48,291][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:31:48,789][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:31:49,287][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:31:49,785][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:31:50,283][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:31:50,782][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:31:51,279][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:31:51,777][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:31:52,277][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:31:52,777][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:31:53,276][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:31:53,775][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:31:54,275][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:31:54,774][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:31:55,274][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:31:55,773][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:31:56,271][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:31:56,769][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:31:57,267][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:31:57,765][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:31:58,264][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:31:58,762][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:31:59,259][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:31:59,758][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:32:00,256][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:32:00,753][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:32:01,252][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:32:01,750][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:32:02,249][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:32:02,746][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:32:03,241][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:32:03,738][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:32:04,237][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:32:04,736][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:32:05,234][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:32:05,733][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:32:06,233][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:32:06,732][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:32:07,231][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:32:07,730][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:32:08,228][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:32:08,725][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:32:09,222][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:32:09,718][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:32:10,215][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:32:10,712][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:32:11,209][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:32:11,707][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:32:12,206][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:32:12,704][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:32:13,201][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:32:13,698][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:32:14,196][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:32:14,693][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:32:15,191][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:32:15,689][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:32:16,187][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:32:16,685][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:32:17,183][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:32:17,681][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:32:18,179][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:32:18,678][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:32:19,176][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:32:19,674][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:32:20,171][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:32:20,669][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:32:21,165][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:32:21,662][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:32:22,159][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:32:22,657][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:32:23,153][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:32:23,651][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:32:24,152][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:32:24,651][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:32:25,151][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:32:25,650][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:32:26,150][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:32:26,650][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:32:27,149][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:32:27,649][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:32:28,147][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:32:28,646][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:32:29,144][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:32:29,642][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:32:30,140][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:32:30,637][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:32:31,134][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:32:31,632][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:32:32,128][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:32:32,626][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:32:33,126][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:32:33,624][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:32:34,121][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:32:34,619][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:32:35,116][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:32:35,614][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:32:36,112][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:32:36,611][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:32:37,109][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:32:37,608][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:32:38,106][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:32:38,603][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:32:39,101][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:32:39,599][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21674 tokens. [2026-03-25 23:32:40,235][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:32:40,967][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:32:40,970][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:32:40,971][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:32:41,625][__main__][INFO] - Iteration 135 took 1m 14s (9.80% Gen, 89.32% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 11m 16s. Estimated total time: 62h 9m 15s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 18s, 500 more iterations: 10h 21m 32s. [2026-03-25 23:32:41,627][__main__][INFO] - Starting iteration 135. [2026-03-25 23:32:42,038][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:32:42,038][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:32:42,617][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:32:49,216][__main__][INFO] - Number of regex retries in iteration 135: 1 [2026-03-25 23:32:49,217][__main__][INFO] - agents played in iteration 135 are Alice, Bob [2026-03-25 23:32:50,094][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:32:50,138][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:32:50,139][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:32:50,139][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:32:50,742][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:32:51,195][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:32:51,698][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:32:52,196][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:32:52,694][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:32:53,192][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:32:53,690][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:32:54,188][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:32:54,687][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:32:55,188][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:32:55,686][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:32:56,185][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:32:56,683][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:32:57,181][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:32:57,680][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:32:58,178][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:32:58,677][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:32:59,175][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:32:59,672][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:33:00,170][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:33:00,668][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:33:01,165][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:33:01,665][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:33:02,163][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:33:02,661][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:33:03,159][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:33:03,657][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:33:04,155][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:33:04,653][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:33:05,151][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:33:05,650][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:33:06,148][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:33:06,647][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:33:07,145][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:33:07,644][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:33:08,141][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:33:08,638][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:33:09,135][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:33:09,633][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:33:10,130][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:33:10,628][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:33:11,127][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:33:11,626][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:33:12,125][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:33:12,623][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:33:13,121][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:33:13,619][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:33:14,117][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:33:14,615][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:33:15,113][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:33:15,611][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:33:16,108][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:33:16,606][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:33:17,103][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:33:17,601][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:33:18,098][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:33:18,599][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:33:19,097][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:33:19,595][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:33:20,093][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:33:20,591][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:33:21,089][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:33:21,586][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:33:22,083][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:33:22,580][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:33:23,077][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:33:23,574][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:33:24,074][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:33:24,571][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:33:25,069][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:33:25,566][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:33:26,064][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:33:26,561][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:33:27,059][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:33:27,557][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:33:28,056][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:33:28,556][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:33:29,055][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:33:29,554][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:33:30,053][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:33:30,552][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:33:31,050][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:33:31,547][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:33:32,047][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:33:32,546][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:33:33,044][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:33:33,541][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:33:34,039][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:33:34,537][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:33:35,034][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:33:35,532][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:33:36,030][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:33:36,529][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:33:37,027][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:33:37,525][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:33:38,023][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:33:38,520][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:33:39,017][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:33:39,514][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:33:40,011][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:33:40,508][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:33:41,001][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:33:41,499][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:33:41,997][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:33:42,494][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:33:42,993][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:33:43,491][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:33:43,989][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:33:44,488][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:33:44,987][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:33:45,487][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:33:45,987][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:33:46,486][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:33:46,984][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:33:47,483][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:33:47,982][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:33:48,482][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:33:48,980][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:33:49,478][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:33:49,977][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:33:50,475][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:33:50,973][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:33:51,472][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:33:51,970][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:33:52,469][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:33:52,967][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:33:53,466][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:33:53,966][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:33:54,464][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21712 tokens. [2026-03-25 23:33:55,093][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 23:33:55,822][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:33:55,824][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:33:55,826][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:33:56,471][__main__][INFO] - Iteration 136 took 1m 14s (9.64% Gen, 89.49% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 59h 2m 27s. Estimated total time: 62h 1m 41s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 3s, 500 more iterations: 10h 20m 16s. [2026-03-25 23:33:56,473][__main__][INFO] - Starting iteration 136. [2026-03-25 23:33:56,875][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:33:56,876][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:34:03,090][__main__][INFO] - Number of regex retries in iteration 136: 0 [2026-03-25 23:34:03,091][__main__][INFO] - agents played in iteration 136 are Alice, Bob [2026-03-25 23:34:03,967][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:34:04,286][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:34:04,287][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:34:04,288][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:34:04,888][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:34:05,343][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:34:05,845][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:34:06,343][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:34:06,840][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:34:07,337][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:34:07,835][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:34:08,332][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:34:08,830][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:34:09,329][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:34:09,828][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:34:10,325][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:34:10,824][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:34:11,323][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:34:11,820][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:34:12,318][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:34:12,815][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:34:13,315][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:34:13,813][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:34:14,311][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:34:14,809][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:34:15,310][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:34:15,811][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:34:16,310][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:34:16,809][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:34:17,326][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:34:17,829][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:34:18,330][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:34:18,830][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:34:19,329][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:34:19,829][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:34:20,328][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:34:20,828][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:34:21,328][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:34:21,828][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:34:22,327][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:34:22,827][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:34:23,326][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:34:23,824][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:34:24,323][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:34:24,821][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:34:25,319][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:34:25,818][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:34:26,315][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:34:26,813][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:34:27,311][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:34:27,809][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:34:28,307][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:34:28,806][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:34:29,307][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:34:29,808][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:34:30,307][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:34:30,806][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:34:31,304][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:34:31,803][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:34:32,300][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:34:32,800][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:34:33,298][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:34:33,797][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:34:34,296][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:34:34,795][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:34:35,294][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:34:35,794][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:34:36,293][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:34:36,793][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:34:37,290][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:34:37,788][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:34:38,286][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:34:38,784][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:34:39,280][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:34:39,778][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:34:40,274][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:34:40,772][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:34:41,270][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:34:41,769][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:34:42,267][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:34:42,766][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:34:43,264][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:34:43,761][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:34:44,258][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:34:44,756][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:34:45,254][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:34:45,752][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:34:46,252][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:34:46,750][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:34:47,248][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:34:47,746][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:34:48,246][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:34:48,745][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:34:49,246][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:34:49,745][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:34:50,246][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:34:50,746][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:34:51,246][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:34:51,744][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:34:52,243][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:34:52,741][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:34:53,240][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:34:53,738][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:34:54,236][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:34:54,735][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:34:55,234][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:34:55,732][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:34:56,231][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:34:56,730][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:34:57,228][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:34:57,727][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:34:58,226][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:34:58,724][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:34:59,221][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:34:59,718][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:35:00,215][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:35:00,712][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:35:01,211][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:35:01,710][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:35:02,209][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:35:02,710][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:35:03,209][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:35:03,707][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:35:04,208][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:35:04,707][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:35:05,205][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:35:05,704][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:35:06,202][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:35:06,702][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:35:07,201][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:35:07,698][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:35:08,196][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:35:08,695][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-25 23:35:09,330][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-25 23:35:10,059][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:35:10,062][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:35:10,063][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:35:10,711][__main__][INFO] - Iteration 137 took 1m 13s (8.42% Gen, 90.70% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 31m 19s. Estimated total time: 61h 31m 48s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 3s, 500 more iterations: 10h 15m 18s. [2026-03-25 23:35:10,713][__main__][INFO] - Starting iteration 137. [2026-03-25 23:35:11,115][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:35:11,116][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:35:17,627][__main__][INFO] - Number of regex retries in iteration 137: 0 [2026-03-25 23:35:17,628][__main__][INFO] - agents played in iteration 137 are Alice, Bob [2026-03-25 23:35:18,509][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:35:18,554][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:35:18,554][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:35:18,555][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:35:19,159][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:35:19,613][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:35:20,117][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:35:20,617][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:35:21,116][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:35:21,616][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:35:22,115][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:35:22,615][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:35:23,114][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:35:23,612][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:35:24,109][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:35:24,609][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:35:25,108][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:35:25,606][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:35:26,104][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:35:26,602][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:35:27,099][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:35:27,597][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:35:28,095][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:35:28,594][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:35:29,092][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:35:29,591][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:35:30,089][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:35:30,587][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:35:31,086][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:35:31,584][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:35:32,082][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:35:32,580][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:35:33,079][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:35:33,577][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:35:34,076][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:35:34,574][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:35:35,073][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:35:35,570][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:35:36,067][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:35:36,563][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:35:37,061][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:35:37,559][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:35:38,055][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:35:38,553][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:35:39,051][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:35:39,549][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:35:40,048][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:35:40,547][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:35:41,047][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:35:41,546][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:35:42,046][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:35:42,545][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:35:43,043][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:35:43,541][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:35:44,039][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:35:44,537][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:35:45,035][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:35:45,537][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:35:46,037][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:35:46,537][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:35:47,035][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:35:47,534][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:35:48,033][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:35:48,533][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:35:49,031][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:35:49,530][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:35:50,028][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:35:50,527][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:35:51,025][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:35:51,525][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:35:52,024][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:35:52,523][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:35:53,022][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:35:53,520][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:35:54,019][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:35:54,519][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:35:55,019][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:35:55,516][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:35:56,013][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:35:56,510][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:35:57,008][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:35:57,507][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:35:58,007][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:35:58,505][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:35:59,002][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:35:59,500][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:35:59,998][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:36:00,496][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:36:00,994][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:36:01,492][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:36:01,991][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:36:02,489][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:36:02,988][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:36:03,487][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:36:03,987][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:36:04,485][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:36:04,982][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:36:05,480][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:36:05,978][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:36:06,477][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:36:06,975][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:36:07,473][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:36:07,971][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:36:08,468][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:36:08,965][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:36:09,461][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:36:09,958][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:36:10,456][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:36:10,953][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:36:11,451][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:36:11,949][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:36:12,448][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:36:12,946][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:36:13,444][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:36:13,941][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:36:14,439][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:36:14,937][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:36:15,435][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:36:15,934][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:36:16,432][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:36:16,931][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:36:17,429][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:36:17,928][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:36:18,429][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:36:18,929][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:36:19,428][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:36:19,926][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:36:20,424][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:36:20,922][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:36:21,420][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:36:21,918][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:36:22,415][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:36:22,913][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21742 tokens. [2026-03-25 23:36:23,553][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-25 23:36:24,290][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:36:24,292][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:36:24,294][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:36:24,945][__main__][INFO] - Iteration 138 took 1m 13s (8.82% Gen, 90.30% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 29m 48s. Estimated total time: 61h 31m 31s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 3s, 500 more iterations: 10h 15m 15s. [2026-03-25 23:36:24,947][__main__][INFO] - Starting iteration 138. [2026-03-25 23:36:25,346][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:36:25,347][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:36:31,812][__main__][INFO] - Number of regex retries in iteration 138: 0 [2026-03-25 23:36:31,813][__main__][INFO] - agents played in iteration 138 are Alice, Bob [2026-03-25 23:36:32,684][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:36:32,729][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:36:32,729][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:36:32,730][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:36:33,330][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:36:33,786][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:36:34,289][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:36:34,788][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:36:35,286][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:36:35,784][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:36:36,282][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:36:36,781][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:36:37,280][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:36:37,778][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:36:38,277][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:36:38,777][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:36:39,275][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:36:39,773][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:36:40,271][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:36:40,771][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:36:41,270][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:36:41,768][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:36:42,265][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:36:42,762][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:36:43,261][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:36:43,758][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:36:44,256][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:36:44,753][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:36:45,251][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:36:45,750][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:36:46,248][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:36:46,748][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:36:47,246][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:36:47,744][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:36:48,243][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:36:48,742][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:36:49,240][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:36:49,738][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:36:50,236][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:36:50,734][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:36:51,232][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:36:51,732][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:36:52,230][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:36:52,729][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:36:53,227][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:36:53,726][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:36:54,224][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:36:54,722][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:36:55,220][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:36:55,718][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:36:56,215][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:36:56,713][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:36:57,211][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:36:57,712][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:36:58,211][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:36:58,709][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:36:59,209][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:36:59,708][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:37:00,207][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:37:00,705][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:37:01,203][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:37:01,701][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:37:02,198][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:37:02,695][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:37:03,193][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:37:03,691][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:37:04,188][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:37:04,685][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:37:05,183][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:37:05,682][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:37:06,180][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:37:06,678][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:37:07,177][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:37:07,675][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:37:08,174][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:37:08,672][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:37:09,171][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:37:09,669][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:37:10,169][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:37:10,668][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:37:11,167][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:37:11,666][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:37:12,164][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:37:12,663][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:37:13,161][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:37:13,658][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:37:14,156][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:37:14,654][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:37:15,153][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:37:15,654][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:37:16,153][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:37:16,651][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:37:17,150][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:37:17,650][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:37:18,149][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:37:18,650][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:37:19,149][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:37:19,649][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:37:20,147][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:37:20,645][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:37:21,143][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:37:21,641][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:37:22,139][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:37:22,637][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:37:23,136][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:37:23,634][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:37:24,131][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:37:24,630][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:37:25,129][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:37:25,627][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:37:26,125][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:37:26,623][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:37:27,121][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:37:27,620][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:37:28,117][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:37:28,615][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:37:29,113][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:37:29,611][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:37:30,111][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:37:30,610][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:37:31,109][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:37:31,608][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:37:32,107][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:37:32,606][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:37:33,105][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:37:33,602][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:37:34,100][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:37:34,598][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:37:35,096][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:37:35,596][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:37:36,094][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:37:36,593][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:37:37,091][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-25 23:37:37,718][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 23:37:38,452][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:37:38,454][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:37:38,456][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:37:39,309][__main__][INFO] - Iteration 139 took 1m 13s (8.74% Gen, 90.10% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 35m 15s. Estimated total time: 61h 38m 12s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 16s, 500 more iterations: 10h 16m 22s. [2026-03-25 23:37:39,312][__main__][INFO] - Starting iteration 139. [2026-03-25 23:37:39,714][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:37:39,714][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:37:46,141][__main__][INFO] - Number of regex retries in iteration 139: 0 [2026-03-25 23:37:46,142][__main__][INFO] - agents played in iteration 139 are Alice, Bob [2026-03-25 23:37:47,020][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:37:47,342][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:37:47,343][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:37:47,343][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:37:47,951][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:37:48,405][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:37:48,908][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:37:49,405][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:37:49,903][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:37:50,402][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:37:50,899][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:37:51,396][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:37:51,894][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:37:52,394][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:37:52,893][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:37:53,390][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:37:53,888][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:37:54,386][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:37:54,883][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:37:55,381][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:37:55,878][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:37:56,378][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:37:56,877][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:37:57,375][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:37:57,874][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:37:58,373][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:37:58,871][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:37:59,370][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:37:59,870][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:38:00,387][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:38:00,891][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:38:01,390][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:38:01,889][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:38:02,388][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:38:02,888][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:38:03,387][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:38:03,887][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:38:04,385][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:38:04,883][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:38:05,380][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:38:05,879][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:38:06,379][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:38:06,878][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:38:07,376][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:38:07,875][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:38:08,374][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:38:08,872][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:38:09,371][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:38:09,869][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:38:10,369][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:38:10,869][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:38:11,368][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:38:11,867][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:38:12,367][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:38:12,865][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:38:13,364][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:38:13,862][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:38:14,361][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:38:14,860][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:38:15,360][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:38:15,858][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:38:16,356][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:38:16,854][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:38:17,353][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:38:17,851][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:38:18,352][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:38:18,852][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:38:19,351][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:38:19,849][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:38:20,347][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:38:20,844][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:38:21,342][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:38:21,839][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:38:22,336][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:38:22,835][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:38:23,333][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:38:23,831][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:38:24,329][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:38:24,827][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:38:25,326][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:38:25,823][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:38:26,321][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:38:26,819][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:38:27,317][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:38:27,815][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:38:28,314][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:38:28,813][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:38:29,311][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:38:29,810][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:38:30,309][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:38:30,809][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:38:31,310][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:38:31,808][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:38:32,310][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:38:32,810][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:38:33,309][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:38:33,809][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:38:34,309][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:38:34,809][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:38:35,310][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:38:35,810][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:38:36,309][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:38:36,807][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:38:37,305][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:38:37,802][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:38:38,300][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:38:38,799][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:38:39,298][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:38:39,796][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:38:40,295][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:38:40,793][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:38:41,292][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:38:41,791][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:38:42,290][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:38:42,789][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:38:43,288][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:38:43,786][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:38:44,285][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:38:44,783][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:38:45,282][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:38:45,783][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:38:46,283][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:38:46,783][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:38:47,281][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:38:47,780][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:38:48,281][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:38:48,779][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:38:49,277][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:38:49,776][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:38:50,275][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:38:50,774][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:38:51,273][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:38:51,772][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21737 tokens. [2026-03-25 23:38:52,412][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:38:53,147][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:38:53,149][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:38:53,151][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:38:54,020][__main__][INFO] - Iteration 140 took 1m 14s (8.65% Gen, 90.18% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 58h 51m 7s. Estimated total time: 61h 55m 19s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 50s, 500 more iterations: 10h 19m 13s. [2026-03-25 23:38:54,022][__main__][INFO] - Starting iteration 140. [2026-03-25 23:38:54,422][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:38:54,423][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:38:55,493][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:38:56,612][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:39:00,638][__main__][INFO] - Number of regex retries in iteration 140: 2 [2026-03-25 23:39:00,639][__main__][INFO] - agents played in iteration 140 are Alice, Bob [2026-03-25 23:39:01,513][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:39:01,558][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:39:01,559][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:39:01,559][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:39:02,450][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:39:02,903][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:39:03,404][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:39:03,902][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:39:04,399][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:39:04,897][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:39:05,395][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:39:05,892][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:39:06,391][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:39:06,895][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:39:07,397][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:39:07,899][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:39:08,401][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:39:08,902][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:39:09,403][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:39:09,904][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:39:10,406][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:39:10,903][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:39:11,402][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:39:11,899][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:39:12,398][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:39:12,896][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:39:13,395][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:39:13,894][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:39:14,392][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:39:14,890][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:39:15,388][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:39:15,887][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:39:16,386][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:39:16,885][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:39:17,382][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:39:17,881][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:39:18,380][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:39:18,881][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:39:19,381][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:39:19,881][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:39:20,380][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:39:20,881][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:39:21,381][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:39:21,881][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:39:22,381][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:39:22,880][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:39:23,378][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:39:23,877][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:39:24,376][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:39:24,875][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:39:25,374][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:39:25,872][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:39:26,370][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:39:26,870][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:39:27,371][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:39:27,871][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:39:28,370][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:39:28,870][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:39:29,369][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:39:29,867][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:39:30,366][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:39:30,865][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:39:31,364][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:39:31,862][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:39:32,361][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:39:32,861][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:39:33,359][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:39:33,857][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:39:34,356][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:39:34,854][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:39:35,352][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:39:35,850][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:39:36,347][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:39:36,845][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:39:37,342][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:39:37,840][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:39:38,337][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:39:38,838][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:39:39,339][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:39:39,840][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:39:40,342][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:39:40,843][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:39:41,344][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:39:41,845][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:39:42,349][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:39:42,850][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:39:43,349][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:39:43,848][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:39:44,347][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:39:44,845][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:39:45,343][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:39:45,841][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:39:46,339][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:39:46,837][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:39:47,332][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:39:47,831][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:39:48,331][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:39:48,830][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:39:49,329][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:39:49,828][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:39:50,326][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:39:50,826][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:39:51,325][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:39:51,824][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:39:52,323][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:39:52,822][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:39:53,321][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:39:53,819][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:39:54,318][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:39:54,817][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:39:55,316][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:39:55,814][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:39:56,312][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:39:56,811][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:39:57,309][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:39:57,806][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:39:58,304][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:39:58,803][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:39:59,301][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:39:59,801][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:40:00,299][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:40:00,797][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:40:01,296][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:40:01,795][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:40:02,294][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:40:02,792][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:40:03,290][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:40:03,789][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:40:04,289][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:40:04,786][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:40:05,283][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:40:05,781][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:40:06,279][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21727 tokens. [2026-03-25 23:40:06,916][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-25 23:40:07,640][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:40:07,643][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:40:07,645][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:40:08,392][__main__][INFO] - Iteration 141 took 1m 13s (8.40% Gen, 90.58% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 58h 33m 7s. Estimated total time: 61h 38m 33s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 17s, 500 more iterations: 10h 16m 25s. [2026-03-25 23:40:08,394][__main__][INFO] - Starting iteration 141. [2026-03-25 23:40:08,798][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:40:08,798][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:40:15,118][__main__][INFO] - Number of regex retries in iteration 141: 0 [2026-03-25 23:40:15,118][__main__][INFO] - agents played in iteration 141 are Alice, Bob [2026-03-25 23:40:15,989][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:40:16,034][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:40:16,034][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:40:16,035][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:40:16,645][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:40:17,096][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:40:17,596][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:40:18,091][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:40:18,587][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:40:19,081][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:40:19,577][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:40:20,073][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:40:20,568][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:40:21,068][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:40:21,567][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:40:22,066][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:40:22,565][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:40:23,064][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:40:23,562][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:40:24,061][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:40:24,559][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:40:25,059][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:40:25,559][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:40:26,059][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:40:26,558][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:40:27,057][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:40:27,556][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:40:28,057][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:40:28,556][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:40:29,055][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:40:29,553][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:40:30,051][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:40:30,548][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:40:31,045][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:40:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:40:32,039][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:40:32,537][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:40:33,036][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:40:33,535][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:40:34,034][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:40:34,532][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:40:35,030][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:40:35,528][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:40:36,026][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:40:36,524][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:40:37,022][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:40:37,520][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:40:38,018][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:40:38,516][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:40:39,014][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:40:39,512][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:40:40,010][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:40:40,508][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:40:41,027][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:40:41,531][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:40:42,030][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:40:42,530][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:40:43,030][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:40:43,531][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:40:44,031][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:40:44,532][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:40:45,029][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:40:45,527][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:40:46,024][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:40:46,522][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:40:47,019][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:40:47,517][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:40:48,017][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:40:48,517][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:40:49,013][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:40:49,508][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:40:50,002][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:40:50,496][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:40:50,991][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:40:51,485][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:40:51,979][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:40:52,475][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:40:52,974][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:40:53,472][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:40:53,971][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:40:54,470][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:40:54,971][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:40:55,469][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:40:55,968][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:40:56,465][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:40:56,965][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:40:57,464][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:40:57,964][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:40:58,463][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:40:58,963][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:40:59,461][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:40:59,960][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:41:00,459][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:41:00,957][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:41:01,455][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:41:01,952][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:41:02,450][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:41:02,949][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:41:03,446][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:41:03,944][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:41:04,440][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:41:04,940][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:41:05,438][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:41:05,938][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:41:06,437][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:41:06,935][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:41:07,434][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:41:07,932][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:41:08,430][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:41:08,931][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:41:09,431][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:41:09,930][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:41:10,429][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:41:10,926][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:41:11,423][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:41:11,921][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:41:12,418][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:41:12,917][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:41:13,416][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:41:13,915][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:41:14,414][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:41:14,914][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:41:15,414][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:41:15,913][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:41:16,412][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:41:16,911][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:41:17,409][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:41:17,908][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:41:18,408][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:41:18,908][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:41:19,406][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:41:19,905][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:41:20,403][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-25 23:41:21,036][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.27%, ΔTime: 00:01:04 [2026-03-25 23:41:21,770][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:41:21,772][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:41:21,774][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:41:22,425][__main__][INFO] - Iteration 142 took 1m 13s (8.58% Gen, 90.53% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 14m 42s. Estimated total time: 61h 21m 23s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 42s, 500 more iterations: 10h 13m 33s. [2026-03-25 23:41:22,427][__main__][INFO] - Starting iteration 142. [2026-03-25 23:41:22,825][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:41:22,826][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:41:23,407][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:41:28,499][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:41:29,554][__main__][INFO] - Number of regex retries in iteration 142: 2 [2026-03-25 23:41:29,555][__main__][INFO] - agents played in iteration 142 are Alice, Bob [2026-03-25 23:41:30,433][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:41:30,477][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:41:30,478][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:41:30,478][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:41:31,082][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:41:31,539][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:41:32,041][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:41:32,540][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:41:33,039][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:41:33,537][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:41:34,036][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:41:34,535][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:41:35,034][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:41:35,534][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:41:36,033][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:41:36,532][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:41:37,031][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:41:37,529][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:41:38,027][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:41:38,525][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:41:39,023][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:41:39,520][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:41:40,018][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:41:40,516][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:41:41,013][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:41:41,510][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:41:42,007][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:41:42,504][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:41:43,002][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:41:43,501][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:41:43,998][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:41:44,496][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:41:44,995][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:41:45,495][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:41:45,995][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:41:46,493][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:41:46,990][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:41:47,487][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:41:47,984][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:41:48,484][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:41:48,981][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:41:49,478][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:41:49,977][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:41:50,475][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:41:50,973][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:41:51,471][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:41:51,969][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:41:52,466][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:41:52,964][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:41:53,462][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:41:53,961][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:41:54,459][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:41:54,958][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:41:55,456][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:41:55,956][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:41:56,455][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:41:56,955][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:41:57,453][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:41:57,951][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:41:58,450][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:41:58,950][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:41:59,446][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:41:59,942][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:42:00,439][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:42:00,935][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:42:01,431][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:42:01,930][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:42:02,428][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:42:02,924][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:42:03,422][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:42:03,920][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:42:04,419][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:42:04,917][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:42:05,416][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:42:05,914][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:42:06,414][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:42:06,913][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:42:07,412][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:42:07,910][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:42:08,408][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:42:08,907][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:42:09,405][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:42:09,903][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:42:10,403][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:42:10,902][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:42:11,400][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:42:11,897][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:42:12,394][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:42:12,891][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:42:13,389][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:42:13,885][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:42:14,382][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:42:14,878][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:42:15,376][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:42:15,874][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:42:16,372][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:42:16,870][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:42:17,367][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:42:17,865][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:42:18,362][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:42:18,861][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:42:19,359][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:42:19,857][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:42:20,354][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:42:20,851][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:42:21,350][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:42:21,847][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:42:22,344][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:42:22,840][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:42:23,338][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:42:23,836][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:42:24,334][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:42:24,832][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:42:25,331][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:42:25,829][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:42:26,325][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:42:26,824][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:42:27,321][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:42:27,818][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:42:28,316][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:42:28,814][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:42:29,313][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:42:29,811][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:42:30,309][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:42:30,807][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:42:31,305][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:42:31,801][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:42:32,298][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:42:32,795][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:42:33,293][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:42:33,791][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:42:34,288][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:42:34,785][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21744 tokens. [2026-03-25 23:42:35,413][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 23:42:36,147][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:42:36,149][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:42:36,151][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:42:36,806][__main__][INFO] - Iteration 143 took 1m 13s (9.10% Gen, 90.02% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 31m 9s. Estimated total time: 61h 39m 4s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 18s, 500 more iterations: 10h 16m 30s. [2026-03-25 23:42:36,808][__main__][INFO] - Starting iteration 143. [2026-03-25 23:42:37,210][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:42:37,210][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:42:40,477][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:42:43,595][__main__][INFO] - Number of regex retries in iteration 143: 1 [2026-03-25 23:42:43,596][__main__][INFO] - agents played in iteration 143 are Alice, Bob [2026-03-25 23:42:44,478][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:42:44,522][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:42:44,523][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:42:44,523][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:42:45,137][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:42:45,593][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:42:46,098][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:42:46,599][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:42:47,099][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:42:47,599][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:42:48,100][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:42:48,601][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:42:49,101][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:42:49,599][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:42:50,097][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:42:50,596][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:42:51,095][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:42:51,594][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:42:52,091][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:42:52,589][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:42:53,087][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:42:53,583][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:42:54,081][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:42:54,578][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:42:55,077][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:42:55,574][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:42:56,072][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:42:56,569][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:42:57,067][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:42:57,567][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:42:58,067][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:42:58,567][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:42:59,067][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:42:59,567][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:43:00,068][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:43:00,567][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:43:01,066][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:43:01,566][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:43:02,066][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:43:02,565][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:43:03,065][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:43:03,564][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:43:04,064][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:43:04,563][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:43:05,062][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:43:05,560][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:43:06,059][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:43:06,558][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:43:07,058][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:43:07,557][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:43:08,057][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:43:08,556][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:43:09,055][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:43:09,554][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:43:10,053][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:43:10,552][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:43:11,051][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:43:11,551][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:43:12,052][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:43:12,551][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:43:13,052][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:43:13,551][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:43:14,049][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:43:14,548][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:43:15,046][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:43:15,545][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:43:16,045][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:43:16,543][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:43:17,042][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:43:17,542][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:43:18,043][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:43:18,544][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:43:19,044][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:43:19,542][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:43:20,041][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:43:20,541][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:43:21,040][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:43:21,539][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:43:22,037][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:43:22,535][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:43:23,034][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:43:23,532][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:43:24,030][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:43:24,530][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:43:25,030][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:43:25,528][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:43:26,025][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:43:26,522][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:43:27,019][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:43:27,516][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:43:28,013][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:43:28,512][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:43:29,010][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:43:29,512][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:43:30,011][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:43:30,510][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:43:31,009][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:43:31,508][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:43:32,006][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:43:32,506][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:43:33,004][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:43:33,503][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:43:34,002][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:43:34,501][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:43:35,001][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:43:35,499][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:43:35,998][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:43:36,497][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:43:36,996][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:43:37,496][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:43:37,995][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:43:38,494][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:43:38,993][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:43:39,492][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:43:39,990][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:43:40,490][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:43:40,990][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:43:41,489][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:43:41,987][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:43:42,485][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:43:42,982][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:43:43,481][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:43:43,979][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:43:44,477][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:43:44,976][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:43:45,474][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:43:45,972][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:43:46,470][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:43:46,968][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:43:47,466][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:43:47,964][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:43:48,463][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:43:48,961][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21738 tokens. [2026-03-25 23:43:49,597][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 23:43:50,333][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:43:50,335][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:43:50,337][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:43:51,009][__main__][INFO] - Iteration 144 took 1m 13s (8.65% Gen, 90.43% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 20m 52s. Estimated total time: 61h 30m 1s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 0s, 500 more iterations: 10h 15m 0s. [2026-03-25 23:43:51,011][__main__][INFO] - Starting iteration 144. [2026-03-25 23:43:51,414][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:43:51,414][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:43:57,730][__main__][INFO] - Number of regex retries in iteration 144: 0 [2026-03-25 23:43:57,731][__main__][INFO] - agents played in iteration 144 are Alice, Bob [2026-03-25 23:43:58,605][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:43:58,650][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:43:58,650][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:43:58,651][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:43:59,257][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:43:59,710][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:44:00,212][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:44:00,709][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:44:01,207][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:44:01,703][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:44:02,200][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:44:02,697][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:44:03,194][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:44:03,693][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:44:04,191][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:44:04,690][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:44:05,188][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:44:05,686][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:44:06,184][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:44:06,682][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:44:07,180][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:44:07,677][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:44:08,174][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:44:08,671][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:44:09,169][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:44:09,665][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:44:10,162][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:44:10,659][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:44:11,157][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:44:11,657][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:44:12,157][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:44:12,657][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:44:13,156][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:44:13,656][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:44:14,155][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:44:14,655][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:44:15,153][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:44:15,651][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:44:16,150][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:44:16,648][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:44:17,146][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:44:17,644][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:44:18,143][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:44:18,643][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:44:19,142][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:44:19,642][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:44:20,139][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:44:20,637][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:44:21,136][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:44:21,634][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:44:22,133][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:44:22,629][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:44:23,129][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:44:23,627][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:44:24,125][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:44:24,622][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:44:25,120][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:44:25,617][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:44:26,114][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:44:26,611][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:44:27,110][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:44:27,608][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:44:28,104][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:44:28,601][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:44:29,098][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:44:29,597][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:44:30,093][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:44:30,591][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:44:31,087][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:44:31,583][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:44:32,079][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:44:32,577][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:44:33,073][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:44:33,570][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:44:34,067][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:44:34,564][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:44:35,061][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:44:35,559][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:44:36,057][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:44:36,557][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:44:37,056][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:44:37,554][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:44:38,053][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:44:38,551][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:44:39,049][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:44:39,546][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:44:40,043][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:44:40,539][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:44:41,036][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:44:41,534][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:44:42,031][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:44:42,528][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:44:43,025][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:44:43,524][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:44:44,023][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:44:44,521][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:44:45,019][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:44:45,520][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:44:46,022][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:44:46,521][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:44:47,019][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:44:47,517][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:44:48,016][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:44:48,515][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:44:49,013][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:44:49,511][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:44:50,010][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:44:50,508][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:44:51,006][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:44:51,503][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:44:52,001][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:44:52,499][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:44:52,997][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:44:53,494][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:44:53,992][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:44:54,491][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:44:54,989][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:44:55,487][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:44:55,985][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:44:56,482][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:44:56,980][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:44:57,477][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:44:57,975][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:44:58,473][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:44:58,971][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:44:59,469][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:44:59,965][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:45:00,463][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:45:00,960][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:45:01,457][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:45:01,954][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:45:02,452][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:45:02,950][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21733 tokens. [2026-03-25 23:45:03,585][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-25 23:45:04,328][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:45:04,330][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:45:04,331][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:45:04,976][__main__][INFO] - Iteration 145 took 1m 13s (8.59% Gen, 90.53% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 7m 45s. Estimated total time: 61h 18m 8s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 36s, 500 more iterations: 10h 13m 1s. [2026-03-25 23:45:04,978][__main__][INFO] - Starting iteration 145. [2026-03-25 23:45:05,378][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:45:05,379][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:45:05,964][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:45:05,969][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:45:10,853][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:45:10,891][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:45:11,929][__main__][INFO] - Number of regex retries in iteration 145: 4 [2026-03-25 23:45:11,930][__main__][INFO] - agents played in iteration 145 are Alice, Bob [2026-03-25 23:45:12,805][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:45:12,849][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:45:12,850][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:45:12,851][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:45:13,452][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:45:13,908][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:45:14,412][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:45:14,912][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:45:15,413][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:45:15,913][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:45:16,413][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:45:16,912][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:45:17,413][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:45:17,911][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:45:18,411][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:45:18,913][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:45:19,412][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:45:19,911][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:45:20,411][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:45:20,910][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:45:21,410][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:45:21,908][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:45:22,408][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:45:22,907][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:45:23,405][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:45:23,904][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:45:24,403][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:45:24,902][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:45:25,401][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:45:25,900][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:45:26,398][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:45:26,897][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:45:27,395][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:45:27,894][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:45:28,392][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:45:28,890][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:45:29,388][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:45:29,886][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:45:30,384][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:45:30,882][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:45:31,380][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:45:31,880][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:45:32,377][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:45:32,876][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:45:33,373][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:45:33,872][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:45:34,371][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:45:34,872][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:45:35,372][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:45:35,872][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:45:36,372][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:45:36,871][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:45:37,370][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:45:37,868][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:45:38,366][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:45:38,866][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:45:39,364][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:45:39,863][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:45:40,361][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:45:40,859][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:45:41,358][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:45:41,857][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:45:42,356][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:45:42,854][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:45:43,353][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:45:43,852][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:45:44,352][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:45:44,850][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:45:45,349][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:45:45,849][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:45:46,349][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:45:46,851][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:45:47,351][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:45:47,852][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:45:48,355][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:45:48,856][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:45:49,356][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:45:49,854][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:45:50,353][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:45:50,852][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:45:51,352][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:45:51,852][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:45:52,349][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:45:52,847][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:45:53,346][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:45:53,845][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:45:54,344][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:45:54,843][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:45:55,342][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:45:55,841][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:45:56,339][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:45:56,838][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:45:57,338][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:45:57,837][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:45:58,336][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:45:58,834][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:45:59,333][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:45:59,832][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:46:00,332][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:46:00,832][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:46:01,331][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:46:01,832][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:46:02,331][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:46:02,830][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:46:03,329][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:46:03,827][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:46:04,324][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:46:04,822][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:46:05,320][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:46:05,819][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:46:06,320][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:46:06,819][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:46:07,318][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:46:07,817][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:46:08,316][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:46:08,815][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:46:09,315][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:46:09,813][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:46:10,312][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:46:10,811][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:46:11,309][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:46:11,807][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:46:12,305][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:46:12,803][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:46:13,301][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:46:13,799][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:46:14,298][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:46:14,797][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:46:15,295][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:46:15,793][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:46:16,292][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:46:16,790][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:46:17,288][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21718 tokens. [2026-03-25 23:46:17,918][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-25 23:46:18,646][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:46:18,649][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:46:18,650][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:46:19,303][__main__][INFO] - Iteration 146 took 1m 13s (8.86% Gen, 90.25% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 24m 38s. Estimated total time: 61h 36m 16s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 12s, 500 more iterations: 10h 16m 2s. [2026-03-25 23:46:19,306][__main__][INFO] - Starting iteration 146. [2026-03-25 23:46:19,716][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:46:19,716][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:46:25,936][__main__][INFO] - Number of regex retries in iteration 146: 0 [2026-03-25 23:46:25,937][__main__][INFO] - agents played in iteration 146 are Alice, Bob [2026-03-25 23:46:26,815][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:46:26,859][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:46:26,860][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:46:26,860][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:46:27,466][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:46:27,922][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:46:28,425][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:46:28,924][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:46:29,422][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:46:29,920][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:46:30,419][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:46:30,917][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:46:31,416][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:46:31,915][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:46:32,413][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:46:32,913][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:46:33,412][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:46:33,911][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:46:34,409][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:46:34,908][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:46:35,406][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:46:35,902][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:46:36,399][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:46:36,897][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:46:37,394][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:46:37,891][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:46:38,389][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:46:38,886][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:46:39,383][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:46:39,882][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:46:40,381][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:46:40,879][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:46:41,378][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:46:41,877][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:46:42,375][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:46:42,875][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:46:43,372][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:46:43,869][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:46:44,366][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:46:44,862][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:46:45,360][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:46:45,858][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:46:46,356][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:46:46,853][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:46:47,350][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:46:47,845][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:46:48,342][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:46:48,838][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:46:49,334][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:46:49,829][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:46:50,324][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:46:50,819][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:46:51,315][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:46:51,814][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:46:52,312][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:46:52,810][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:46:53,309][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:46:53,807][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:46:54,304][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:46:54,802][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:46:55,300][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:46:55,798][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:46:56,296][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:46:56,794][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:46:57,292][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:46:57,789][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:46:58,287][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:46:58,784][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:46:59,281][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:46:59,779][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:47:00,278][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:47:00,776][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:47:01,274][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:47:01,772][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:47:02,270][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:47:02,768][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:47:03,267][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:47:03,766][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:47:04,264][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:47:04,763][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:47:05,261][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:47:05,760][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:47:06,259][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:47:06,757][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:47:07,256][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:47:07,754][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:47:08,253][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:47:08,751][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:47:09,250][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:47:09,748][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:47:10,245][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:47:10,742][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:47:11,240][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:47:11,738][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:47:12,236][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:47:12,735][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:47:13,234][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:47:13,733][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:47:14,232][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:47:14,730][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:47:15,227][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:47:15,724][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:47:16,221][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:47:16,718][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:47:17,216][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:47:17,714][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:47:18,212][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:47:18,709][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:47:19,206][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:47:19,701][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:47:20,196][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:47:20,691][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:47:21,186][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:47:21,681][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:47:22,176][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:47:22,673][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:47:23,168][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:47:23,667][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:47:24,166][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:47:24,664][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:47:25,162][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:47:25,660][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:47:26,159][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:47:26,657][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:47:27,156][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:47:27,654][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:47:28,152][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:47:28,651][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:47:29,148][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:47:29,646][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:47:30,144][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:47:30,641][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:47:31,139][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-25 23:47:31,773][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-25 23:47:32,507][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:47:32,510][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:47:32,511][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:47:33,158][__main__][INFO] - Iteration 147 took 1m 13s (8.47% Gen, 90.65% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 59m 22s. Estimated total time: 61h 12m 13s. Time estimates for 10 more iterations: 12m 14s, 100 more iterations: 2h 2m 24s, 500 more iterations: 10h 12m 2s. [2026-03-25 23:47:33,160][__main__][INFO] - Starting iteration 147. [2026-03-25 23:47:33,559][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:47:33,560][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:47:39,841][__main__][INFO] - Number of regex retries in iteration 147: 0 [2026-03-25 23:47:39,842][__main__][INFO] - agents played in iteration 147 are Alice, Bob [2026-03-25 23:47:40,713][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:47:40,757][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:47:40,757][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:47:40,758][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:47:41,364][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:47:41,819][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:47:42,321][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:47:42,818][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:47:43,316][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:47:43,815][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:47:44,314][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:47:44,812][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:47:45,312][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:47:45,810][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:47:46,308][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:47:46,806][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:47:47,303][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:47:47,801][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:47:48,302][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:47:48,801][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:47:49,300][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:47:49,798][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:47:50,297][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:47:50,795][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:47:51,294][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:47:51,793][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:47:52,292][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:47:52,791][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:47:53,291][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:47:53,788][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:47:54,286][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:47:54,784][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:47:55,283][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:47:55,781][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:47:56,278][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:47:56,776][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:47:57,275][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:47:57,773][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:47:58,272][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:47:58,769][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:47:59,267][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:47:59,765][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:48:00,263][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:48:00,761][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:48:01,259][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:48:01,757][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:48:02,256][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:48:02,754][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:48:03,253][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:48:03,752][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:48:04,251][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:48:04,752][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:48:05,250][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:48:05,747][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:48:06,245][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:48:06,743][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:48:07,241][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:48:07,738][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:48:08,236][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:48:08,734][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:48:09,233][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:48:09,732][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:48:10,230][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:48:10,728][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:48:11,226][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:48:11,723][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:48:12,220][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:48:12,719][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:48:13,218][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:48:13,716][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:48:14,213][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:48:14,711][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:48:15,210][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:48:15,708][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:48:16,205][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:48:16,703][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:48:17,202][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:48:17,699][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:48:18,198][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:48:18,696][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:48:19,193][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:48:19,748][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:48:20,246][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:48:20,743][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:48:21,241][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:48:21,738][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:48:22,235][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:48:22,733][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:48:23,233][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:48:23,730][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:48:24,227][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:48:24,724][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:48:25,224][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:48:25,721][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:48:26,219][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:48:26,717][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:48:27,215][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:48:27,713][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:48:28,212][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:48:28,710][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:48:29,208][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:48:29,706][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:48:30,204][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:48:30,702][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:48:31,201][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:48:31,699][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:48:32,198][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:48:32,696][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:48:33,194][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:48:33,692][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:48:34,189][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:48:34,687][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:48:35,184][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:48:35,683][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:48:36,181][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:48:36,679][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:48:37,178][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:48:37,676][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:48:38,173][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:48:38,670][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:48:39,167][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:48:39,664][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:48:40,162][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:48:40,661][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:48:41,159][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:48:41,657][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:48:42,155][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:48:42,654][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:48:43,154][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:48:43,653][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:48:44,152][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:48:44,653][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:48:45,151][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21737 tokens. [2026-03-25 23:48:45,782][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 23:48:46,518][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:48:46,520][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:48:46,522][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:48:47,364][__main__][INFO] - Iteration 148 took 1m 13s (8.51% Gen, 90.34% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 16m 12s. Estimated total time: 61h 30m 17s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 0s, 500 more iterations: 10h 15m 2s. [2026-03-25 23:48:47,367][__main__][INFO] - Starting iteration 148. [2026-03-25 23:48:47,768][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:48:47,768][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:48:49,003][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:48:54,709][__main__][INFO] - Number of regex retries in iteration 148: 1 [2026-03-25 23:48:54,710][__main__][INFO] - agents played in iteration 148 are Alice, Bob [2026-03-25 23:48:55,588][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:48:55,633][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:48:55,634][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:48:55,634][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:48:56,243][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:48:56,699][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:48:57,202][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:48:57,702][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:48:58,201][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:48:58,701][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:48:59,201][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:48:59,701][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:49:00,201][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:49:00,700][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:49:01,198][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:49:01,697][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:49:02,196][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:49:02,694][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:49:03,194][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:49:03,694][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:49:04,194][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:49:04,693][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:49:05,193][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:49:05,691][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:49:06,190][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:49:06,689][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:49:07,187][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:49:07,686][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:49:08,184][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:49:08,685][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:49:09,183][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:49:09,682][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:49:10,181][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:49:10,679][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:49:11,179][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:49:11,678][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:49:12,177][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:49:12,676][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:49:13,175][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:49:13,674][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:49:14,173][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:49:14,673][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:49:15,172][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:49:15,675][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:49:16,176][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:49:16,675][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:49:17,173][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:49:17,672][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:49:18,173][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:49:18,673][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:49:19,172][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:49:19,672][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:49:20,170][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:49:20,668][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:49:21,166][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:49:21,664][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:49:22,164][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:49:22,662][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:49:23,162][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:49:23,660][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:49:24,159][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:49:24,658][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:49:25,157][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:49:25,657][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:49:26,156][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:49:26,656][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:49:27,154][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:49:27,653][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:49:28,152][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:49:28,652][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:49:29,152][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:49:29,651][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:49:30,151][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:49:30,650][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:49:31,149][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:49:31,647][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:49:32,146][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:49:32,644][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:49:33,143][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:49:33,642][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:49:34,142][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:49:34,642][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:49:35,140][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:49:35,638][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:49:36,138][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:49:36,637][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:49:37,136][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:49:37,636][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:49:38,134][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:49:38,632][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:49:39,130][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:49:39,628][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:49:40,125][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:49:40,624][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:49:41,121][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:49:41,619][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:49:42,118][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:49:42,617][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:49:43,115][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:49:43,614][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:49:44,112][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:49:44,610][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:49:45,108][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:49:45,606][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:49:46,103][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:49:46,602][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:49:47,102][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:49:47,601][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:49:48,101][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:49:48,600][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:49:49,098][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:49:49,597][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:49:50,095][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:49:50,594][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:49:51,092][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:49:51,590][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:49:52,088][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:49:52,586][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:49:53,084][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:49:53,582][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:49:54,082][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:49:54,581][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:49:55,081][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:49:55,579][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:49:56,079][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:49:56,577][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:49:57,076][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:49:57,575][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:49:58,075][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:49:58,573][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:49:59,072][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:49:59,571][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:50:00,070][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21718 tokens. [2026-03-25 23:50:00,712][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-25 23:50:01,450][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:50:01,452][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:50:01,454][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:50:02,262][__main__][INFO] - Iteration 149 took 1m 14s (9.32% Gen, 89.59% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 49m 26s. Estimated total time: 62h 4m 47s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 47s. [2026-03-25 23:50:02,265][__main__][INFO] - Starting iteration 149. [2026-03-25 23:50:02,663][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:50:02,664][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:50:05,517][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:50:09,059][__main__][INFO] - Number of regex retries in iteration 149: 1 [2026-03-25 23:50:09,060][__main__][INFO] - agents played in iteration 149 are Alice, Bob [2026-03-25 23:50:09,931][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:50:09,975][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:50:09,976][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:50:09,976][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:50:10,591][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:50:11,046][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:50:11,548][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:50:12,047][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:50:12,544][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:50:13,043][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:50:13,542][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:50:14,040][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:50:14,540][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:50:15,036][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:50:15,532][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:50:16,028][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:50:16,524][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:50:17,020][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:50:17,515][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:50:18,012][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:50:18,509][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:50:19,009][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:50:19,507][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:50:20,005][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:50:20,504][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:50:21,001][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:50:21,500][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:50:21,999][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:50:22,496][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:50:22,994][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:50:23,493][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:50:23,992][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:50:24,490][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:50:24,988][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:50:25,486][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:50:25,983][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:50:26,481][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:50:26,981][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:50:27,481][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:50:27,981][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:50:28,480][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:50:28,978][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:50:29,477][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:50:29,976][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:50:30,474][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:50:30,974][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:50:31,473][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:50:31,972][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:50:32,470][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:50:32,968][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:50:33,467][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:50:33,965][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:50:34,462][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:50:34,961][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:50:35,460][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:50:35,959][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:50:36,457][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:50:36,954][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:50:37,453][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:50:37,951][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:50:38,449][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:50:38,947][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:50:39,444][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:50:39,942][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:50:40,439][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:50:40,936][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:50:41,434][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:50:41,930][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:50:42,428][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:50:42,925][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:50:43,423][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:50:43,920][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:50:44,419][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:50:44,917][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:50:45,418][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:50:45,918][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:50:46,417][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:50:46,911][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:50:47,407][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:50:47,903][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:50:48,401][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:50:48,897][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:50:49,394][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:50:49,890][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:50:50,385][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:50:50,885][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:50:51,383][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:50:51,882][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:50:52,381][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:50:52,880][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:50:53,379][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:50:53,878][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:50:54,377][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:50:54,875][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:50:55,373][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:50:55,871][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:50:56,368][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:50:56,865][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:50:57,363][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:50:57,862][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:50:58,360][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:50:58,860][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:50:59,360][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:50:59,859][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:51:00,358][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:51:00,857][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:51:01,356][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:51:01,855][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:51:02,354][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:51:02,853][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:51:03,351][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:51:03,848][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:51:04,346][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:51:04,844][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:51:05,342][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:51:05,841][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:51:06,340][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:51:06,837][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:51:07,335][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:51:07,833][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:51:08,331][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:51:08,828][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:51:09,326][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:51:09,823][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:51:10,321][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:51:10,820][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:51:11,319][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:51:11,818][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:51:12,317][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:51:12,815][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:51:13,315][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:51:13,814][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:51:14,314][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-25 23:51:14,957][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:51:15,692][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:51:15,694][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:51:15,696][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:51:16,355][__main__][INFO] - Iteration 150 took 1m 13s (8.68% Gen, 90.42% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 8m 5s. Estimated total time: 61h 24m 39s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 49s, 500 more iterations: 10h 14m 6s. [2026-03-25 23:51:16,358][__main__][INFO] - Starting iteration 150. [2026-03-25 23:51:16,760][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 2 and human policies 1. [2026-03-25 23:51:16,761][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:51:19,423][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:51:20,526][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:51:23,338][__main__][INFO] - Number of regex retries in iteration 150: 2 [2026-03-25 23:51:23,338][__main__][INFO] - agents played in iteration 150 are Alice, Bob [2026-03-25 23:51:24,223][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:51:24,267][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:51:24,267][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:51:24,268][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:51:24,875][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:51:25,329][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:51:25,831][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:51:26,329][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:51:26,827][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:51:27,325][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:51:27,823][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:51:28,320][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:51:28,819][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:51:29,318][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:51:29,817][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:51:30,314][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:51:30,813][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:51:31,312][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:51:31,810][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:51:32,308][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:51:32,806][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:51:33,301][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:51:33,797][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:51:34,292][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:51:34,788][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:51:35,284][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:51:35,780][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:51:36,277][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:51:36,775][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:51:37,275][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:51:37,773][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:51:38,271][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:51:38,770][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:51:39,268][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:51:39,765][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:51:40,263][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:51:40,763][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:51:41,262][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:51:41,761][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:51:42,259][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:51:42,757][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:51:43,255][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:51:43,754][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:51:44,271][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:51:44,768][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:51:45,266][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:51:45,764][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:51:46,268][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:51:46,766][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:51:47,264][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:51:47,762][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:51:48,262][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:51:48,761][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:51:49,260][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:51:49,757][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:51:50,254][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:51:50,752][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:51:51,251][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:51:51,748][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:51:52,245][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:51:52,742][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:51:53,239][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:51:53,737][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:51:54,234][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:51:54,731][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:51:55,228][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:51:55,725][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:51:56,222][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:51:56,720][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:51:57,218][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:51:57,716][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:51:58,214][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:51:58,713][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:51:59,211][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:51:59,708][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:52:00,205][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:52:00,702][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:52:01,201][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:52:01,700][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:52:02,198][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:52:02,697][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:52:03,195][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:52:03,694][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:52:04,192][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:52:04,689][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:52:05,185][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:52:05,682][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:52:06,178][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:52:06,675][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:52:07,172][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:52:07,669][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:52:08,165][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:52:08,661][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:52:09,160][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:52:09,659][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:52:10,158][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:52:10,656][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:52:11,155][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:52:11,654][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:52:12,153][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:52:12,651][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:52:13,149][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:52:13,647][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:52:14,145][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:52:14,643][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:52:15,142][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:52:15,644][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:52:16,144][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:52:16,642][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:52:17,141][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:52:17,640][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:52:18,139][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:52:18,639][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:52:19,138][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:52:19,636][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:52:20,135][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:52:20,633][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:52:21,131][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:52:21,628][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:52:22,125][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:52:22,623][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:52:23,134][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:52:23,633][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:52:24,131][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:52:24,628][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:52:25,126][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:52:25,624][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:52:26,122][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:52:26,621][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:52:27,120][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:52:27,618][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:52:28,116][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:52:28,615][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-25 23:52:29,256][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-25 23:52:29,889][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:52:29,892][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:52:29,893][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:52:31,119][__main__][INFO] - Iteration 151 took 1m 14s (8.85% Gen, 89.50% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 40m 9s. Estimated total time: 61h 57m 58s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 55s, 500 more iterations: 10h 19m 39s. [2026-03-25 23:52:31,124][__main__][INFO] - Starting iteration 151. [2026-03-25 23:52:31,523][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:52:31,524][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:52:37,769][__main__][INFO] - Number of regex retries in iteration 151: 0 [2026-03-25 23:52:37,770][__main__][INFO] - agents played in iteration 151 are Alice, Bob [2026-03-25 23:52:38,650][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:52:38,694][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:52:38,695][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:52:38,695][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:52:39,298][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:52:39,754][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:52:40,257][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:52:40,756][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:52:41,255][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:52:41,754][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:52:42,252][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:52:42,750][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:52:43,248][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:52:43,747][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:52:44,245][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:52:44,742][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:52:45,241][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:52:45,740][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:52:46,239][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:52:46,737][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:52:47,235][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:52:47,734][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:52:48,235][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:52:48,735][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:52:49,235][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:52:49,734][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:52:50,232][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:52:50,729][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:52:51,227][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:52:51,724][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:52:52,221][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:52:52,718][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:52:53,216][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:52:53,714][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:52:54,211][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:52:54,708][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:52:55,206][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:52:55,706][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:52:56,205][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:52:56,704][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:52:57,204][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:52:57,702][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:52:58,202][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:52:58,701][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:52:59,200][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:52:59,697][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:53:00,197][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:53:00,696][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:53:01,195][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:53:01,695][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:53:02,194][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:53:02,694][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:53:03,192][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:53:03,690][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:53:04,187][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:53:04,685][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:53:05,183][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:53:05,681][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:53:06,178][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:53:06,677][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:53:07,175][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:53:07,672][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:53:08,169][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:53:08,667][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:53:09,164][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:53:09,663][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:53:10,161][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:53:10,659][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:53:11,157][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:53:11,656][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:53:12,154][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:53:12,651][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:53:13,149][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:53:13,646][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:53:14,144][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:53:14,643][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:53:15,141][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:53:15,640][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:53:16,138][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:53:16,637][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:53:17,137][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:53:17,635][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:53:18,136][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:53:18,636][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:53:19,135][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:53:19,633][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:53:20,129][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:53:20,627][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:53:21,125][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:53:21,622][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:53:22,120][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:53:22,618][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:53:23,116][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:53:23,614][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:53:24,111][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:53:24,607][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:53:25,105][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:53:25,601][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:53:26,097][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:53:26,593][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:53:27,089][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:53:27,587][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:53:28,085][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:53:28,584][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:53:29,083][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:53:29,581][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:53:30,079][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:53:30,577][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:53:31,076][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:53:31,576][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:53:32,076][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:53:32,575][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:53:33,073][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:53:33,571][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:53:34,069][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:53:34,567][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:53:35,065][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:53:35,563][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:53:36,060][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:53:36,558][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:53:37,055][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:53:37,554][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:53:38,051][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:53:38,548][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:53:39,045][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:53:39,543][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:53:40,040][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:53:40,538][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:53:41,036][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:53:41,534][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:53:42,032][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:53:42,528][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:53:43,026][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21747 tokens. [2026-03-25 23:53:43,655][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 23:53:44,395][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:53:44,397][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:53:44,398][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:53:45,004][__main__][INFO] - Iteration 152 took 1m 13s (8.50% Gen, 90.67% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 55m 2s. Estimated total time: 61h 14m 5s. Time estimates for 10 more iterations: 12m 14s, 100 more iterations: 2h 2m 28s, 500 more iterations: 10h 12m 20s. [2026-03-25 23:53:45,007][__main__][INFO] - Starting iteration 152. [2026-03-25 23:53:45,405][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:53:45,406][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:53:46,015][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:53:46,021][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:53:52,280][__main__][INFO] - Number of regex retries in iteration 152: 2 [2026-03-25 23:53:52,281][__main__][INFO] - agents played in iteration 152 are Alice, Bob [2026-03-25 23:53:53,162][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:53:53,206][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:53:53,207][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:53:53,207][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:53:53,808][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:53:54,262][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:53:54,763][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:53:55,261][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:53:55,757][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:53:56,255][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:53:56,751][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:53:57,247][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:53:57,744][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:53:58,244][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:53:58,742][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:53:59,241][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:53:59,739][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:54:00,236][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:54:00,734][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:54:01,232][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:54:01,729][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:54:02,227][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:54:02,724][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:54:03,222][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:54:03,720][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:54:04,218][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:54:04,715][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:54:05,212][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:54:05,709][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:54:06,207][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:54:06,705][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:54:07,204][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:54:07,703][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:54:08,201][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:54:08,700][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:54:09,198][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:54:09,695][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:54:10,193][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:54:10,691][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:54:11,188][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:54:11,686][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:54:12,184][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:54:12,683][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:54:13,181][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:54:13,679][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:54:14,176][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:54:14,673][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:54:15,171][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:54:15,669][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:54:16,167][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:54:16,664][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:54:17,163][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:54:17,661][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:54:18,160][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:54:18,659][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:54:19,158][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:54:19,656][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:54:20,155][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:54:20,654][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:54:21,152][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:54:21,651][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:54:22,151][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:54:22,651][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:54:23,149][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:54:23,647][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:54:24,147][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:54:24,646][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:54:25,146][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:54:25,644][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:54:26,140][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:54:26,638][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:54:27,137][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:54:27,635][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:54:28,132][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:54:28,630][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:54:29,126][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:54:29,623][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:54:30,122][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:54:30,620][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:54:31,118][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:54:31,617][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:54:32,116][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:54:32,615][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:54:33,114][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:54:33,612][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:54:34,110][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:54:34,607][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:54:35,105][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:54:35,603][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:54:36,100][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:54:36,598][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:54:37,097][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:54:37,594][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:54:38,093][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:54:38,590][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:54:39,088][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:54:39,585][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:54:40,083][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:54:40,581][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:54:41,080][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:54:41,578][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:54:42,077][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:54:42,576][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:54:43,075][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:54:43,572][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:54:44,071][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:54:44,568][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:54:45,066][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:54:45,565][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:54:46,064][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:54:46,562][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:54:47,060][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:54:47,559][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:54:48,058][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:54:48,559][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:54:49,058][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:54:49,556][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:54:50,055][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:54:50,554][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:54:51,051][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:54:51,549][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:54:52,046][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:54:52,543][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:54:53,041][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:54:53,539][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:54:54,038][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:54:54,537][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:54:55,036][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:54:55,535][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:54:56,033][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:54:56,531][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:54:57,029][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:54:57,527][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-25 23:54:58,159][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 23:54:58,890][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:54:58,892][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:54:58,894][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:54:59,764][__main__][INFO] - Iteration 153 took 1m 14s (9.24% Gen, 89.58% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 37m 39s. Estimated total time: 61h 57m 57s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 55s, 500 more iterations: 10h 19m 39s. [2026-03-25 23:54:59,766][__main__][INFO] - Starting iteration 153. [2026-03-25 23:55:00,168][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:55:00,169][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:55:00,753][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:55:02,990][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:55:06,856][__main__][INFO] - Number of regex retries in iteration 153: 2 [2026-03-25 23:55:06,856][__main__][INFO] - agents played in iteration 153 are Alice, Bob [2026-03-25 23:55:07,732][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:55:07,776][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:55:07,777][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:55:07,777][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:55:08,386][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:55:08,839][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:55:09,342][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:55:09,840][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:55:10,338][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:55:10,837][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:55:11,336][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:55:11,835][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:55:12,334][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:55:12,835][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:55:13,333][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:55:13,831][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:55:14,329][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:55:14,826][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:55:15,326][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:55:15,828][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:55:16,326][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:55:16,823][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:55:17,321][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:55:17,819][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:55:18,319][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:55:18,817][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:55:19,315][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:55:19,813][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:55:20,311][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:55:20,805][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:55:21,299][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:55:21,794][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:55:22,288][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:55:22,782][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:55:23,276][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:55:23,770][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:55:24,264][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:55:24,763][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:55:25,264][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:55:25,763][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:55:26,262][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:55:26,761][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:55:27,260][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:55:27,759][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:55:28,258][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:55:28,756][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:55:29,254][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:55:29,752][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:55:30,249][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:55:30,746][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:55:31,243][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:55:31,740][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:55:32,238][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:55:32,737][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:55:33,237][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:55:33,736][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:55:34,235][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:55:34,735][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:55:35,233][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:55:35,732][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:55:36,230][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:55:36,728][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:55:37,226][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:55:37,724][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:55:38,222][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:55:38,720][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:55:39,218][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:55:39,716][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:55:40,213][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:55:40,710][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:55:41,207][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:55:41,704][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:55:42,201][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:55:42,699][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:55:43,196][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:55:43,694][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:55:44,191][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:55:44,689][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:55:45,187][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:55:45,685][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:55:46,184][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:55:46,682][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:55:47,181][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:55:47,679][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:55:48,178][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:55:48,677][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:55:49,176][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:55:49,674][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:55:50,175][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:55:50,674][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:55:51,173][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:55:51,672][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:55:52,171][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:55:52,666][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:55:53,160][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:55:53,656][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:55:54,149][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:55:54,643][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:55:55,138][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:55:55,633][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:55:56,126][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:55:56,627][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:55:57,126][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:55:57,624][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:55:58,123][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:55:58,622][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:55:59,121][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:55:59,619][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:56:00,119][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:56:00,616][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:56:01,112][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:56:01,608][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:56:02,104][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:56:02,601][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:56:03,097][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:56:03,594][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:56:04,089][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:56:04,587][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:56:05,085][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:56:05,584][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:56:06,083][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:56:06,582][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:56:07,081][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:56:07,579][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:56:08,078][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:56:08,576][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:56:09,077][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:56:09,576][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:56:10,075][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:56:10,575][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:56:11,072][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:56:11,570][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:56:12,067][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-25 23:56:12,705][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-25 23:56:13,442][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:56:13,444][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:56:13,445][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:56:14,284][__main__][INFO] - Iteration 154 took 1m 14s (9.02% Gen, 89.84% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 24m 17s. Estimated total time: 61h 45m 49s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 31s, 500 more iterations: 10h 17m 38s. [2026-03-25 23:56:14,286][__main__][INFO] - Starting iteration 154. [2026-03-25 23:56:14,686][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:56:14,687][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:56:15,275][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:56:15,812][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:56:21,723][__main__][INFO] - Number of regex retries in iteration 154: 2 [2026-03-25 23:56:21,724][__main__][INFO] - agents played in iteration 154 are Alice, Bob [2026-03-25 23:56:22,607][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:56:22,651][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:56:22,652][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:56:22,652][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:56:23,274][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:56:23,728][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:56:24,231][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:56:24,729][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:56:25,228][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:56:25,725][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:56:26,224][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:56:26,724][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:56:27,223][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:56:27,723][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:56:28,222][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:56:28,721][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:56:29,220][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:56:29,719][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:56:30,217][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:56:30,717][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:56:31,215][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:56:31,712][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:56:32,210][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:56:32,708][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:56:33,205][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:56:33,702][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:56:34,199][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:56:34,697][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:56:35,195][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:56:35,694][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:56:36,193][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:56:36,691][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:56:37,189][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:56:37,687][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:56:38,185][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:56:38,684][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:56:39,182][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:56:39,681][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:56:40,179][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:56:40,679][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:56:41,177][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:56:41,675][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:56:42,173][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:56:42,672][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:56:43,170][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:56:43,668][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:56:44,167][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:56:44,665][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:56:45,163][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:56:45,665][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:56:46,164][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:56:46,663][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:56:47,162][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:56:47,663][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:56:48,163][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:56:48,664][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:56:49,164][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:56:49,664][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:56:50,164][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:56:50,664][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:56:51,165][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:56:51,664][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:56:52,164][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:56:52,663][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:56:53,163][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:56:53,663][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:56:54,163][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:56:54,663][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:56:55,162][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:56:55,661][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:56:56,160][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:56:56,659][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:56:57,158][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:56:57,656][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:56:58,155][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:56:58,654][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:56:59,152][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:56:59,650][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:57:00,149][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:57:00,647][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:57:01,145][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:57:01,644][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:57:02,144][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:57:02,642][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:57:03,141][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:57:03,639][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:57:04,137][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:57:04,636][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:57:05,134][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:57:05,632][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:57:06,129][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:57:06,627][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:57:07,125][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:57:07,623][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:57:08,122][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:57:08,620][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:57:09,118][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:57:09,617][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:57:10,116][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:57:10,616][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:57:11,115][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:57:11,613][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:57:12,112][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:57:12,610][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:57:13,108][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:57:13,607][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:57:14,105][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:57:14,604][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:57:15,102][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:57:15,600][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:57:16,098][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:57:16,596][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:57:17,094][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:57:17,591][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:57:18,090][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:57:18,589][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:57:19,087][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:57:19,586][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:57:20,086][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:57:20,586][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:57:21,085][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:57:21,585][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:57:22,085][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:57:22,586][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:57:23,085][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:57:23,584][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:57:24,084][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:57:24,583][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:57:25,083][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:57:25,582][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:57:26,082][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:57:26,581][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:57:27,080][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21741 tokens. [2026-03-25 23:57:27,732][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-25 23:57:28,549][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:57:28,552][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:57:28,553][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:57:29,211][__main__][INFO] - Iteration 155 took 1m 14s (9.44% Gen, 89.67% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 58h 43m 28s. Estimated total time: 62h 6m 15s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 12s, 500 more iterations: 10h 21m 2s. [2026-03-25 23:57:29,296][__main__][INFO] - Starting iteration 155. [2026-03-25 23:57:29,697][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:57:29,698][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:57:30,278][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:57:31,529][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:57:36,087][__main__][INFO] - Number of regex retries in iteration 155: 2 [2026-03-25 23:57:36,088][__main__][INFO] - agents played in iteration 155 are Alice, Bob [2026-03-25 23:57:36,961][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:57:37,005][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:57:37,006][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:57:37,006][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:57:37,612][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:57:38,068][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:57:38,572][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:57:39,072][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:57:39,570][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:57:40,069][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:57:40,568][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:57:41,067][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:57:41,566][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:57:42,064][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:57:42,561][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:57:43,060][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:57:43,558][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:57:44,057][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:57:44,555][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:57:45,054][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:57:45,551][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:57:46,050][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:57:46,549][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:57:47,047][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:57:47,545][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:57:48,044][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:57:48,545][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:57:49,044][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:57:49,544][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:57:50,043][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:57:50,541][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:57:51,041][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:57:51,540][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:57:52,040][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:57:52,539][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:57:53,038][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:57:53,537][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:57:54,034][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:57:54,531][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:57:55,028][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:57:55,525][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:57:56,022][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:57:56,519][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:57:57,017][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:57:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:57:58,013][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:57:58,510][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:57:59,007][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:57:59,505][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:58:00,002][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:58:00,500][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:58:00,997][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:58:01,496][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:58:01,995][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:58:02,494][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:58:02,991][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:58:03,490][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:58:03,988][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:58:04,486][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:58:04,985][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:58:05,483][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:58:05,982][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:58:06,480][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:58:06,979][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:58:07,477][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:58:07,975][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:58:08,474][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:58:08,971][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:58:09,468][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:58:09,969][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:58:10,468][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:58:10,968][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:58:11,468][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:58:11,968][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:58:12,467][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:58:12,965][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:58:13,464][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:58:13,962][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:58:14,459][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:58:14,957][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:58:15,457][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:58:15,957][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:58:16,454][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:58:16,951][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:58:17,448][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:58:17,947][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:58:18,448][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:58:18,946][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:58:19,445][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:58:19,943][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:58:20,443][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:58:20,942][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:58:21,441][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:58:21,940][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:58:22,439][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:58:22,938][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:58:23,435][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:58:23,933][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:58:24,431][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:58:24,929][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:58:25,427][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:58:25,926][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:58:26,425][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:58:26,922][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:58:27,421][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:58:27,918][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:58:28,418][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:58:28,916][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:58:29,415][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:58:29,914][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:58:30,411][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:58:30,909][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:58:31,406][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:58:31,903][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:58:32,401][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:58:32,899][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:58:33,397][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:58:33,895][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:58:34,394][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:58:34,893][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:58:35,391][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:58:35,890][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:58:36,389][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:58:36,887][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:58:37,385][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:58:37,883][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:58:38,383][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:58:38,882][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:58:39,381][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:58:39,880][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:58:40,378][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:58:40,876][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:58:41,374][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21735 tokens. [2026-03-25 23:58:42,017][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-25 23:58:42,756][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:58:42,758][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:58:42,760][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:58:43,512][__main__][INFO] - Iteration 156 took 1m 13s (8.66% Gen, 90.32% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 6m 45s. Estimated total time: 61h 30m 46s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 1s, 500 more iterations: 10h 15m 7s. [2026-03-25 23:58:43,514][__main__][INFO] - Starting iteration 156. [2026-03-25 23:58:43,920][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:58:43,921][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-25 23:58:44,490][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:58:44,496][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-25 23:58:50,349][__main__][INFO] - Number of regex retries in iteration 156: 2 [2026-03-25 23:58:50,350][__main__][INFO] - agents played in iteration 156 are Alice, Bob [2026-03-25 23:58:51,222][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:58:51,266][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-25 23:58:51,267][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-25 23:58:51,268][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-25 23:58:51,881][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-25 23:58:52,334][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-25 23:58:52,836][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-25 23:58:53,333][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-25 23:58:53,831][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-25 23:58:54,328][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-25 23:58:54,825][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-25 23:58:55,323][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-25 23:58:55,822][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-25 23:58:56,323][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-25 23:58:56,823][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-25 23:58:57,321][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-25 23:58:57,820][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-25 23:58:58,319][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-25 23:58:58,819][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-25 23:58:59,318][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-25 23:58:59,817][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-25 23:59:00,316][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-25 23:59:00,813][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-25 23:59:01,311][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-25 23:59:01,810][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-25 23:59:02,307][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-25 23:59:02,805][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-25 23:59:03,303][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-25 23:59:03,801][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-25 23:59:04,299][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-25 23:59:04,798][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-25 23:59:05,295][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-25 23:59:05,794][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-25 23:59:06,291][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-25 23:59:06,788][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-25 23:59:07,286][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-25 23:59:07,785][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-25 23:59:08,285][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-25 23:59:08,784][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-25 23:59:09,283][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-25 23:59:09,782][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-25 23:59:10,280][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-25 23:59:10,778][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-25 23:59:11,276][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-25 23:59:11,777][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-25 23:59:12,272][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-25 23:59:12,768][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-25 23:59:13,264][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-25 23:59:13,761][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-25 23:59:14,258][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-25 23:59:14,753][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-25 23:59:15,249][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-25 23:59:15,745][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-25 23:59:16,244][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-25 23:59:16,742][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-25 23:59:17,240][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-25 23:59:17,739][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-25 23:59:18,239][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-25 23:59:18,737][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-25 23:59:19,234][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-25 23:59:19,733][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-25 23:59:20,230][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-25 23:59:20,727][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-25 23:59:21,225][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-25 23:59:21,724][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-25 23:59:22,222][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-25 23:59:22,719][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-25 23:59:23,217][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-25 23:59:23,715][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-25 23:59:24,212][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-25 23:59:24,708][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-25 23:59:25,206][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-25 23:59:25,705][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-25 23:59:26,203][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-25 23:59:26,700][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-25 23:59:27,198][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-25 23:59:27,697][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-25 23:59:28,195][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-25 23:59:28,693][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-25 23:59:29,191][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-25 23:59:29,690][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-25 23:59:30,189][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-25 23:59:30,686][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-25 23:59:31,185][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-25 23:59:31,685][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-25 23:59:32,182][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-25 23:59:32,679][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-25 23:59:33,176][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-25 23:59:33,673][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-25 23:59:34,170][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-25 23:59:34,668][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-25 23:59:35,165][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-25 23:59:35,663][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-25 23:59:36,161][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-25 23:59:36,658][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-25 23:59:37,156][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-25 23:59:37,653][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-25 23:59:38,150][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-25 23:59:38,648][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-25 23:59:39,145][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-25 23:59:39,644][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-25 23:59:40,143][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-25 23:59:40,642][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-25 23:59:41,141][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-25 23:59:41,641][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-25 23:59:42,141][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-25 23:59:42,640][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-25 23:59:43,140][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-25 23:59:43,639][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-25 23:59:44,135][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-25 23:59:44,631][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-25 23:59:45,127][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-25 23:59:45,627][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-25 23:59:46,125][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-25 23:59:46,621][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-25 23:59:47,117][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-25 23:59:47,613][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-25 23:59:48,111][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-25 23:59:48,610][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-25 23:59:49,107][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-25 23:59:49,605][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-25 23:59:50,103][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-25 23:59:50,600][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-25 23:59:51,098][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-25 23:59:51,596][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-25 23:59:52,096][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-25 23:59:52,596][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-25 23:59:53,096][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-25 23:59:53,595][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-25 23:59:54,093][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-25 23:59:54,592][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-25 23:59:55,091][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-25 23:59:55,590][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21672 tokens. [2026-03-25 23:59:56,228][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-25 23:59:56,958][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-25 23:59:56,960][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-25 23:59:56,961][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-25 23:59:57,613][__main__][INFO] - Iteration 157 took 1m 13s (8.72% Gen, 90.39% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 59m 26s. Estimated total time: 61h 24m 41s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 49s, 500 more iterations: 10h 14m 6s. [2026-03-25 23:59:57,616][__main__][INFO] - Starting iteration 157. [2026-03-25 23:59:58,016][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-25 23:59:58,016][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:00:04,808][__main__][INFO] - Number of regex retries in iteration 157: 0 [2026-03-26 00:00:04,809][__main__][INFO] - agents played in iteration 157 are Alice, Bob [2026-03-26 00:00:05,683][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:00:05,727][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:00:05,728][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:00:05,729][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:00:06,333][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:00:06,786][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:00:07,288][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:00:07,784][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:00:08,280][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:00:08,775][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:00:09,272][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:00:09,768][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:00:10,264][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:00:10,764][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:00:11,263][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:00:11,762][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:00:12,260][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:00:12,758][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:00:13,257][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:00:13,756][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:00:14,253][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:00:14,751][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:00:15,248][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:00:15,746][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:00:16,244][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:00:16,742][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:00:17,241][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:00:17,739][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:00:18,238][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:00:18,738][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:00:19,236][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:00:19,734][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:00:20,233][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:00:20,731][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:00:21,229][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:00:21,727][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:00:22,225][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:00:22,722][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:00:23,220][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:00:23,717][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:00:24,214][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:00:24,711][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:00:25,208][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:00:25,705][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:00:26,203][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:00:26,717][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:00:27,219][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:00:27,719][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:00:28,219][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:00:28,719][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:00:29,218][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:00:29,718][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:00:30,216][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:00:30,714][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:00:31,211][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:00:31,708][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:00:32,206][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:00:32,703][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:00:33,200][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:00:33,698][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:00:34,194][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:00:34,693][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:00:35,191][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:00:35,690][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:00:36,189][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:00:36,688][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:00:37,186][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:00:37,686][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:00:38,185][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:00:38,681][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:00:39,176][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:00:39,672][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:00:40,167][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:00:40,663][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:00:41,159][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:00:41,656][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:00:42,151][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:00:42,649][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:00:43,147][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:00:43,645][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:00:44,143][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:00:44,641][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:00:45,140][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:00:45,639][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:00:46,139][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:00:46,637][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:00:47,137][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:00:47,636][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:00:48,134][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:00:48,633][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:00:49,130][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:00:49,628][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:00:50,126][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:00:50,625][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:00:51,124][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:00:51,623][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:00:52,121][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:00:52,619][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:00:53,117][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:00:53,615][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:00:54,113][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:00:54,610][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:00:55,108][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:00:55,606][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:00:56,103][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:00:56,600][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:00:57,097][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:00:57,594][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:00:58,091][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:00:58,589][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:00:59,088][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:00:59,587][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:01:00,086][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:01:00,585][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:01:01,084][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:01:01,582][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:01:02,081][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:01:02,579][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:01:03,077][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:01:03,575][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:01:04,072][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:01:04,569][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:01:05,066][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:01:05,563][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:01:06,061][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:01:06,561][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:01:07,060][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:01:07,561][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:01:08,060][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:01:08,559][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:01:09,059][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:01:09,557][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:01:10,057][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 00:01:10,684][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 00:01:11,439][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:01:11,442][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:01:11,443][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:01:12,225][__main__][INFO] - Iteration 158 took 1m 14s (9.15% Gen, 89.79% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 23m 58s. Estimated total time: 61h 50m 28s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 40s, 500 more iterations: 10h 18m 24s. [2026-03-26 00:01:12,227][__main__][INFO] - Starting iteration 158. [2026-03-26 00:01:12,625][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:01:12,625][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:01:13,209][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:01:13,213][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:01:18,433][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:01:19,182][__main__][INFO] - Number of regex retries in iteration 158: 3 [2026-03-26 00:01:19,183][__main__][INFO] - agents played in iteration 158 are Alice, Bob [2026-03-26 00:01:20,055][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:01:20,100][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:01:20,100][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:01:20,101][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:01:20,710][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:01:21,168][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:01:21,669][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:01:22,167][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:01:22,666][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:01:23,164][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:01:23,663][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:01:24,161][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:01:24,660][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:01:25,159][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:01:25,658][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:01:26,157][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:01:26,657][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:01:27,156][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:01:27,655][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:01:28,153][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:01:28,652][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:01:29,151][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:01:29,650][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:01:30,149][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:01:30,647][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:01:31,146][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:01:31,645][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:01:32,145][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:01:32,644][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:01:33,143][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:01:33,641][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:01:34,139][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:01:34,638][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:01:35,138][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:01:35,638][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:01:36,136][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:01:36,636][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:01:37,133][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:01:37,630][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:01:38,127][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:01:38,624][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:01:39,122][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:01:39,620][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:01:40,118][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:01:40,617][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:01:41,111][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:01:41,607][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:01:42,103][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:01:42,599][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:01:43,095][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:01:43,590][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:01:44,086][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:01:44,583][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:01:45,083][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:01:45,583][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:01:46,082][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:01:46,581][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:01:47,079][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:01:47,578][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:01:48,076][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:01:48,577][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:01:49,077][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:01:49,576][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:01:50,075][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:01:50,573][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:01:51,072][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:01:51,570][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:01:52,068][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:01:52,567][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:01:53,067][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:01:53,565][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:01:54,064][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:01:54,563][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:01:55,062][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:01:55,561][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:01:56,059][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:01:56,560][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:01:57,060][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:01:57,558][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:01:58,058][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:01:58,556][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:01:59,054][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:01:59,551][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:02:00,049][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:02:00,548][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:02:01,047][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:02:01,546][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:02:02,045][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:02:02,543][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:02:03,043][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:02:03,542][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:02:04,041][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:02:04,540][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:02:05,039][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:02:05,536][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:02:06,034][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:02:06,531][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:02:07,028][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:02:07,526][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:02:08,023][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:02:08,520][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:02:09,019][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:02:09,518][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:02:10,015][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:02:10,512][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:02:11,010][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:02:11,507][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:02:12,004][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:02:12,501][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:02:12,996][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:02:13,493][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:02:13,988][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:02:14,485][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:02:14,982][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:02:15,478][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:02:15,975][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:02:16,471][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:02:16,970][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:02:17,465][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:02:17,965][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:02:18,467][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:02:18,966][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:02:19,466][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:02:19,961][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:02:20,460][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:02:20,959][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:02:21,459][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:02:21,958][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:02:22,458][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:02:22,957][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:02:23,456][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:02:23,955][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:02:24,453][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21677 tokens. [2026-03-26 00:02:25,082][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 00:02:25,814][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:02:25,816][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:02:25,818][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:02:26,468][__main__][INFO] - Iteration 159 took 1m 13s (8.88% Gen, 90.24% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 4m 26s. Estimated total time: 61h 32m 10s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 21s. [2026-03-26 00:02:26,470][__main__][INFO] - Starting iteration 159. [2026-03-26 00:02:26,869][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:02:26,870][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:02:33,799][__main__][INFO] - Number of regex retries in iteration 159: 0 [2026-03-26 00:02:33,800][__main__][INFO] - agents played in iteration 159 are Alice, Bob [2026-03-26 00:02:34,673][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:02:34,718][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:02:34,718][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:02:34,719][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:02:35,324][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:02:35,778][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:02:36,280][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:02:36,778][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:02:37,276][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:02:37,772][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:02:38,270][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:02:38,767][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:02:39,266][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:02:39,767][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:02:40,267][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:02:40,765][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:02:41,265][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:02:41,764][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:02:42,264][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:02:42,764][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:02:43,265][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:02:43,763][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:02:44,260][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:02:44,758][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:02:45,255][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:02:45,755][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:02:46,253][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:02:46,750][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:02:47,247][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:02:47,745][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:02:48,244][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:02:48,742][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:02:49,239][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:02:49,737][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:02:50,233][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:02:50,729][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:02:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:02:51,725][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:02:52,224][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:02:52,723][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:02:53,222][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:02:53,721][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:02:54,220][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:02:54,719][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:02:55,219][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:02:55,719][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:02:56,219][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:02:56,718][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:02:57,217][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:02:57,716][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:02:58,214][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:02:58,714][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:02:59,213][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:02:59,712][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:03:00,212][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:03:00,711][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:03:01,211][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:03:01,711][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:03:02,211][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:03:02,710][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:03:03,209][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:03:03,708][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:03:04,206][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:03:04,705][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:03:05,203][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:03:05,702][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:03:06,201][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:03:06,700][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:03:07,199][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:03:07,697][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:03:08,195][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:03:08,692][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:03:09,189][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:03:09,687][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:03:10,184][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:03:10,682][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:03:11,180][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:03:11,680][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:03:12,179][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:03:12,679][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:03:13,179][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:03:13,678][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:03:14,177][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:03:14,679][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:03:15,179][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:03:15,677][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:03:16,175][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:03:16,672][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:03:17,169][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:03:17,667][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:03:18,166][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:03:18,665][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:03:19,162][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:03:19,660][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:03:20,158][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:03:20,656][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:03:21,153][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:03:21,650][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:03:22,146][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:03:22,644][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:03:23,141][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:03:23,641][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:03:24,140][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:03:24,639][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:03:25,140][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:03:25,639][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:03:26,138][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:03:26,636][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:03:27,137][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:03:27,638][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:03:28,137][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:03:28,637][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:03:29,136][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:03:29,634][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:03:30,133][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:03:30,631][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:03:31,131][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:03:31,630][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:03:32,130][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:03:32,629][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:03:33,128][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:03:33,627][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:03:34,126][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:03:34,626][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:03:35,126][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:03:35,626][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:03:36,124][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:03:36,622][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:03:37,121][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:03:37,619][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:03:38,117][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:03:38,616][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:03:39,114][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 00:03:39,763][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:03:40,496][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:03:40,498][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:03:40,499][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:03:41,310][__main__][INFO] - Iteration 160 took 1m 14s (9.31% Gen, 89.60% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 33m 4s. Estimated total time: 62h 2m 3s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 4s, 500 more iterations: 10h 20m 20s. [2026-03-26 00:03:41,312][__main__][INFO] - Starting iteration 160. [2026-03-26 00:03:41,709][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:03:41,709][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:03:48,466][__main__][INFO] - Number of regex retries in iteration 160: 0 [2026-03-26 00:03:48,467][__main__][INFO] - agents played in iteration 160 are Alice, Bob [2026-03-26 00:03:49,349][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:03:49,394][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:03:49,394][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:03:49,395][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:03:50,004][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:03:50,461][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:03:50,962][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:03:51,461][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:03:51,958][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:03:52,456][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:03:52,952][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:03:53,450][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:03:53,947][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:03:54,447][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:03:54,946][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:03:55,445][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:03:55,944][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:03:56,442][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:03:56,942][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:03:57,440][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:03:57,939][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:03:58,440][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:03:58,939][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:03:59,438][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:03:59,936][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:04:00,435][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:04:00,933][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:04:01,431][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:04:01,930][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:04:02,427][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:04:02,926][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:04:03,424][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:04:03,922][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:04:04,420][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:04:04,918][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:04:05,416][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:04:05,914][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:04:06,413][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:04:06,910][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:04:07,408][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:04:07,907][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:04:08,405][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:04:08,903][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:04:09,402][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:04:09,900][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:04:10,423][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:04:10,928][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:04:11,429][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:04:11,929][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:04:12,431][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:04:12,931][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:04:13,433][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:04:13,934][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:04:14,433][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:04:14,930][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:04:15,429][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:04:15,929][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:04:16,428][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:04:16,923][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:04:17,422][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:04:17,921][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:04:18,421][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:04:18,918][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:04:19,415][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:04:19,912][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:04:20,410][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:04:20,907][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:04:21,405][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:04:21,903][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:04:22,401][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:04:22,898][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:04:23,395][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:04:23,891][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:04:24,387][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:04:24,884][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:04:25,382][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:04:25,879][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:04:26,378][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:04:26,877][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:04:27,375][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:04:27,874][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:04:28,372][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:04:28,870][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:04:29,368][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:04:29,868][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:04:30,368][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:04:30,866][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:04:31,365][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:04:31,866][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:04:32,365][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:04:32,866][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:04:33,366][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:04:33,865][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:04:34,364][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:04:34,862][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:04:35,361][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:04:35,861][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:04:36,360][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:04:36,862][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:04:37,360][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:04:37,860][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:04:38,359][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:04:38,856][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:04:39,354][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:04:39,851][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:04:40,349][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:04:40,847][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:04:41,346][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:04:41,844][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:04:42,346][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:04:42,847][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:04:43,348][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:04:43,850][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:04:44,351][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:04:44,851][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:04:45,352][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:04:45,853][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:04:46,352][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:04:46,849][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:04:47,347][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:04:47,846][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:04:48,346][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:04:48,846][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:04:49,345][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:04:49,844][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:04:50,342][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:04:50,840][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:04:51,338][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:04:51,835][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:04:52,333][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:04:52,830][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:04:53,327][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:04:53,825][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21723 tokens. [2026-03-26 00:04:54,450][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.56%, ΔTime: 00:01:04 [2026-03-26 00:04:55,180][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:04:55,182][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:04:55,184][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:04:55,836][__main__][INFO] - Iteration 161 took 1m 14s (9.12% Gen, 90.00% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 16m 9s. Estimated total time: 61h 46m 23s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 32s, 500 more iterations: 10h 17m 43s. [2026-03-26 00:04:55,838][__main__][INFO] - Starting iteration 161. [2026-03-26 00:04:56,236][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:04:56,236][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:05:02,784][__main__][INFO] - Number of regex retries in iteration 161: 0 [2026-03-26 00:05:02,785][__main__][INFO] - agents played in iteration 161 are Alice, Bob [2026-03-26 00:05:03,660][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:05:03,704][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:05:03,705][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:05:03,705][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:05:04,314][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:05:04,770][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:05:05,274][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:05:05,774][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:05:06,274][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:05:06,774][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:05:07,274][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:05:07,774][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:05:08,273][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:05:08,771][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:05:09,268][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:05:09,766][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:05:10,264][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:05:10,762][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:05:11,260][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:05:11,757][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:05:12,255][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:05:12,753][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:05:13,250][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:05:13,748][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:05:14,246][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:05:14,745][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:05:15,243][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:05:15,740][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:05:16,238][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:05:16,737][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:05:17,234][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:05:17,733][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:05:18,232][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:05:18,731][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:05:19,229][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:05:19,728][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:05:20,226][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:05:20,724][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:05:21,222][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:05:21,720][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:05:22,221][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:05:22,719][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:05:23,217][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:05:23,715][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:05:24,213][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:05:24,711][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:05:25,208][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:05:25,704][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:05:26,202][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:05:26,700][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:05:27,198][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:05:27,695][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:05:28,192][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:05:28,690][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:05:29,187][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:05:29,685][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:05:30,183][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:05:30,680][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:05:31,179][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:05:31,677][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:05:32,175][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:05:32,674][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:05:33,173][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:05:33,672][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:05:34,171][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:05:34,670][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:05:35,168][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:05:35,668][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:05:36,168][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:05:36,668][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:05:37,167][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:05:37,667][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:05:38,167][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:05:38,667][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:05:39,167][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:05:39,667][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:05:40,167][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:05:40,665][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:05:41,163][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:05:41,662][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:05:42,161][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:05:42,659][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:05:43,158][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:05:43,655][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:05:44,153][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:05:44,651][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:05:45,149][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:05:45,650][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:05:46,149][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:05:46,647][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:05:47,145][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:05:47,644][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:05:48,142][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:05:48,643][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:05:49,141][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:05:49,641][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:05:50,141][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:05:50,640][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:05:51,139][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:05:51,638][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:05:52,137][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:05:52,634][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:05:53,132][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:05:53,630][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:05:54,128][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:05:54,626][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:05:55,124][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:05:55,623][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:05:56,121][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:05:56,619][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:05:57,116][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:05:57,615][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:05:58,113][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:05:58,610][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:05:59,108][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:05:59,605][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:06:00,103][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:06:00,601][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:06:01,098][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:06:01,596][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:06:02,094][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:06:02,591][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:06:03,088][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:06:03,586][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:06:04,084][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:06:04,583][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:06:05,082][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:06:05,581][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:06:06,081][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:06:06,580][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:06:07,078][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:06:07,577][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:06:08,077][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:06:08,707][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:06:09,448][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:06:09,451][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:06:09,452][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:06:10,109][__main__][INFO] - Iteration 162 took 1m 13s (8.86% Gen, 90.25% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 2m 13s. Estimated total time: 61h 33m 41s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 7s, 500 more iterations: 10h 15m 36s. [2026-03-26 00:06:10,111][__main__][INFO] - Starting iteration 162. [2026-03-26 00:06:10,509][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:06:10,509][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:06:16,601][__main__][INFO] - Number of regex retries in iteration 162: 0 [2026-03-26 00:06:16,601][__main__][INFO] - agents played in iteration 162 are Alice, Bob [2026-03-26 00:06:17,478][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:06:17,522][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:06:17,523][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:06:17,523][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:06:18,138][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:06:18,596][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:06:19,101][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:06:19,603][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:06:20,104][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:06:20,606][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:06:21,108][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:06:21,610][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:06:22,112][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:06:22,611][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:06:23,108][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:06:23,607][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:06:24,106][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:06:24,604][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:06:25,102][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:06:25,601][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:06:26,100][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:06:26,598][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:06:27,096][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:06:27,594][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:06:28,093][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:06:28,591][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:06:29,090][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:06:29,589][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:06:30,087][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:06:30,585][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:06:31,083][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:06:31,581][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:06:32,081][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:06:32,579][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:06:33,078][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:06:33,576][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:06:34,074][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:06:34,571][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:06:35,069][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:06:35,567][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:06:36,065][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:06:36,563][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:06:37,062][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:06:37,560][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:06:38,060][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:06:38,558][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:06:39,057][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:06:39,554][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:06:40,052][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:06:40,550][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:06:41,049][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:06:41,548][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:06:42,046][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:06:42,544][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:06:43,043][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:06:43,542][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:06:44,041][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:06:44,540][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:06:45,039][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:06:45,537][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:06:46,036][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:06:46,533][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:06:47,031][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:06:47,528][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:06:48,026][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:06:48,526][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:06:49,025][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:06:49,523][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:06:50,020][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:06:50,520][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:06:51,022][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:06:51,523][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:06:52,024][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:06:52,525][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:06:53,027][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:06:53,528][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:06:54,030][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:06:54,528][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:06:55,026][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:06:55,524][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:06:56,022][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:06:56,521][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:06:57,019][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:06:57,518][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:06:58,016][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:06:58,515][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:06:59,014][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:06:59,512][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:07:00,010][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:07:00,509][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:07:01,008][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:07:01,507][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:07:02,006][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:07:02,503][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:07:03,002][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:07:03,499][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:07:03,998][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:07:04,497][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:07:04,994][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:07:05,492][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:07:05,989][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:07:06,488][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:07:06,986][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:07:07,485][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:07:07,983][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:07:08,481][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:07:08,980][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:07:09,477][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:07:09,975][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:07:10,474][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:07:10,973][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:07:11,471][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:07:11,969][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:07:12,469][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:07:12,969][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:07:13,468][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:07:13,966][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:07:14,465][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:07:14,964][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:07:15,466][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:07:15,966][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:07:16,465][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:07:16,964][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:07:17,463][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:07:17,962][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:07:18,463][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:07:18,963][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:07:19,462][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:07:19,960][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:07:20,459][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:07:20,958][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:07:21,456][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:07:21,954][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21760 tokens. [2026-03-26 00:07:22,582][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 00:07:23,313][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:07:23,316][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:07:23,317][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:07:23,991][__main__][INFO] - Iteration 163 took 1m 13s (8.29% Gen, 90.79% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 41m 27s. Estimated total time: 61h 14m 9s. Time estimates for 10 more iterations: 12m 14s, 100 more iterations: 2h 2m 28s, 500 more iterations: 10h 12m 21s. [2026-03-26 00:07:23,993][__main__][INFO] - Starting iteration 163. [2026-03-26 00:07:24,399][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:07:24,400][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:07:30,730][__main__][INFO] - Number of regex retries in iteration 163: 0 [2026-03-26 00:07:30,731][__main__][INFO] - agents played in iteration 163 are Alice, Bob [2026-03-26 00:07:31,601][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:07:31,646][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:07:31,646][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:07:31,647][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:07:32,251][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:07:32,705][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:07:33,207][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:07:33,706][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:07:34,204][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:07:34,702][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:07:35,203][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:07:35,702][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:07:36,202][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:07:36,702][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:07:37,201][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:07:37,699][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:07:38,197][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:07:38,696][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:07:39,193][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:07:39,692][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:07:40,190][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:07:40,689][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:07:41,187][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:07:41,686][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:07:42,184][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:07:42,681][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:07:43,178][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:07:43,676][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:07:44,174][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:07:44,671][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:07:45,170][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:07:45,668][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:07:46,166][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:07:46,664][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:07:47,161][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:07:47,659][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:07:48,158][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:07:48,658][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:07:49,156][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:07:49,653][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:07:50,152][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:07:50,649][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:07:51,147][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:07:51,645][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:07:52,144][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:07:52,642][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:07:53,141][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:07:53,640][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:07:54,138][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:07:54,636][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:07:55,135][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:07:55,633][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:07:56,130][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:07:56,630][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:07:57,129][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:07:57,628][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:07:58,128][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:07:58,626][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:07:59,125][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:07:59,624][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:08:00,123][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:08:00,621][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:08:01,121][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:08:01,620][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:08:02,119][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:08:02,618][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:08:03,116][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:08:03,615][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:08:04,114][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:08:04,611][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:08:05,109][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:08:05,607][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:08:06,106][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:08:06,604][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:08:07,102][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:08:07,600][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:08:08,098][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:08:08,596][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:08:09,094][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:08:09,592][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:08:10,090][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:08:10,589][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:08:11,087][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:08:11,586][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:08:12,085][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:08:12,583][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:08:13,082][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:08:13,581][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:08:14,080][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:08:14,579][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:08:15,077][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:08:15,575][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:08:16,073][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:08:16,570][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:08:17,068][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:08:17,567][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:08:18,066][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:08:18,566][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:08:19,064][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:08:19,563][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:08:20,061][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:08:20,560][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:08:21,058][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:08:21,556][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:08:22,054][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:08:22,552][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:08:23,049][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:08:23,547][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:08:24,045][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:08:24,543][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:08:25,043][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:08:25,541][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:08:26,042][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:08:26,542][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:08:27,040][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:08:27,539][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:08:28,038][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:08:28,536][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:08:29,035][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:08:29,533][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:08:30,033][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:08:30,531][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:08:31,031][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:08:31,530][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:08:32,036][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:08:32,536][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:08:33,035][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:08:33,534][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:08:34,033][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:08:34,532][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:08:35,031][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:08:35,530][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:08:36,028][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 00:08:36,673][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:08:37,405][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:08:37,407][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:08:37,409][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:08:38,059][__main__][INFO] - Iteration 164 took 1m 13s (8.59% Gen, 90.52% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 49m 5s. Estimated total time: 61h 23m 1s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 46s, 500 more iterations: 10h 13m 50s. [2026-03-26 00:08:38,061][__main__][INFO] - Starting iteration 164. [2026-03-26 00:08:38,460][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:08:38,461][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:08:45,204][__main__][INFO] - Number of regex retries in iteration 164: 0 [2026-03-26 00:08:45,205][__main__][INFO] - agents played in iteration 164 are Alice, Bob [2026-03-26 00:08:46,077][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:08:46,121][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:08:46,122][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:08:46,122][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:08:46,744][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:08:47,205][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:08:47,709][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:08:48,209][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:08:48,709][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:08:49,207][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:08:49,706][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:08:50,204][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:08:50,702][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:08:51,203][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:08:51,702][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:08:52,200][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:08:52,699][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:08:53,199][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:08:53,697][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:08:54,195][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:08:54,693][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:08:55,194][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:08:55,693][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:08:56,192][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:08:56,690][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:08:57,189][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:08:57,688][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:08:58,187][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:08:58,686][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:08:59,185][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:08:59,682][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:09:00,179][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:09:00,677][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:09:01,174][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:09:01,671][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:09:02,169][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:09:02,666][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:09:03,164][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:09:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:09:04,161][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:09:04,659][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:09:05,157][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:09:05,655][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:09:06,153][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:09:06,650][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:09:07,150][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:09:07,648][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:09:08,147][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:09:08,645][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:09:09,144][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:09:09,642][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:09:10,143][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:09:10,642][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:09:11,159][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:09:11,664][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:09:12,164][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:09:12,663][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:09:13,163][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:09:13,662][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:09:14,162][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:09:14,661][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:09:15,159][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:09:15,657][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:09:16,155][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:09:16,653][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:09:17,150][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:09:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:09:18,148][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:09:18,647][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:09:19,145][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:09:19,644][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:09:20,142][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:09:20,640][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:09:21,139][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:09:21,637][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:09:22,135][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:09:22,632][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:09:23,132][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:09:23,630][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:09:24,127][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:09:24,625][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:09:25,123][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:09:25,622][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:09:26,121][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:09:26,619][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:09:27,117][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:09:27,615][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:09:28,113][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:09:28,611][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:09:29,110][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:09:29,608][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:09:30,107][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:09:30,605][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:09:31,102][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:09:31,600][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:09:32,098][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:09:32,594][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:09:33,091][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:09:33,589][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:09:34,088][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:09:34,585][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:09:35,084][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:09:35,582][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:09:36,081][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:09:36,578][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:09:37,076][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:09:37,574][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:09:38,072][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:09:38,570][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:09:39,069][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:09:39,568][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:09:40,067][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:09:40,566][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:09:41,064][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:09:41,563][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:09:42,063][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:09:42,562][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:09:43,061][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:09:43,561][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:09:44,059][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:09:44,557][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:09:45,055][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:09:45,553][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:09:46,050][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:09:46,549][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:09:47,047][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:09:47,547][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:09:48,046][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:09:48,546][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:09:49,046][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:09:49,546][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:09:50,046][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:09:50,545][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 00:09:51,196][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 00:09:51,934][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:09:51,936][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:09:51,938][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:09:52,653][__main__][INFO] - Iteration 165 took 1m 14s (9.09% Gen, 89.95% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 14m 28s. Estimated total time: 61h 49m 38s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 16s. [2026-03-26 00:09:52,655][__main__][INFO] - Starting iteration 165. [2026-03-26 00:09:53,052][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:09:53,053][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:09:53,640][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:09:59,415][__main__][INFO] - Number of regex retries in iteration 165: 1 [2026-03-26 00:09:59,416][__main__][INFO] - agents played in iteration 165 are Alice, Bob [2026-03-26 00:10:00,290][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:10:00,335][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:10:00,335][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:10:00,336][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:10:01,212][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:10:01,666][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:10:02,170][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:10:02,669][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:10:03,168][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:10:03,668][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:10:04,168][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:10:04,668][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:10:05,167][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:10:05,667][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:10:06,164][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:10:06,662][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:10:07,161][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:10:07,659][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:10:08,157][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:10:08,654][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:10:09,153][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:10:09,653][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:10:10,154][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:10:10,653][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:10:11,153][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:10:11,652][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:10:12,150][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:10:12,650][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:10:13,150][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:10:13,648][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:10:14,147][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:10:14,645][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:10:15,144][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:10:15,646][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:10:16,146][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:10:16,644][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:10:17,143][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:10:17,641][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:10:18,140][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:10:18,638][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:10:19,137][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:10:19,634][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:10:20,131][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:10:20,629][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:10:21,128][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:10:21,625][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:10:22,124][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:10:22,621][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:10:23,120][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:10:23,618][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:10:24,116][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:10:24,614][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:10:25,113][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:10:25,612][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:10:26,112][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:10:26,611][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:10:27,110][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:10:27,609][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:10:28,109][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:10:28,608][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:10:29,108][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:10:29,604][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:10:30,101][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:10:30,599][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:10:31,096][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:10:31,593][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:10:32,090][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:10:32,587][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:10:33,085][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:10:33,584][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:10:34,083][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:10:34,581][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:10:35,079][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:10:35,577][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:10:36,075][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:10:36,573][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:10:37,071][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:10:37,568][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:10:38,066][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:10:38,564][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:10:39,062][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:10:39,559][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:10:40,056][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:10:40,553][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:10:41,050][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:10:41,549][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:10:42,048][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:10:42,549][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:10:43,051][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:10:43,551][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:10:44,051][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:10:44,550][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:10:45,049][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:10:45,548][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:10:46,047][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:10:46,545][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:10:47,044][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:10:47,541][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:10:48,039][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:10:48,539][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:10:49,037][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:10:49,534][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:10:50,031][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:10:50,529][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:10:51,027][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:10:51,525][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:10:52,022][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:10:52,520][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:10:53,018][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:10:53,516][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:10:54,021][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:10:54,521][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:10:55,021][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:10:55,520][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:10:56,019][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:10:56,519][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:10:57,017][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:10:57,517][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:10:58,017][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:10:58,517][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:10:59,016][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:10:59,516][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:11:00,018][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:11:00,522][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:11:01,024][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:11:01,524][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:11:02,024][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:11:02,522][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:11:03,021][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:11:03,519][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:11:04,017][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:11:04,515][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:11:05,013][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 00:11:05,656][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:11:06,393][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:11:06,396][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:11:06,398][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:11:07,112][__main__][INFO] - Iteration 166 took 1m 14s (8.59% Gen, 90.44% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 6m 35s. Estimated total time: 61h 43m 0s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 26s, 500 more iterations: 10h 17m 10s. [2026-03-26 00:11:07,114][__main__][INFO] - Starting iteration 166. [2026-03-26 00:11:07,546][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:11:07,547][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:11:13,607][__main__][INFO] - Number of regex retries in iteration 166: 0 [2026-03-26 00:11:13,607][__main__][INFO] - agents played in iteration 166 are Alice, Bob [2026-03-26 00:11:14,482][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:11:14,526][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:11:14,527][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:11:14,528][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:11:15,408][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:11:15,862][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:11:16,365][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:11:16,864][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:11:17,363][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:11:17,861][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:11:18,362][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:11:18,861][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:11:19,361][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:11:19,859][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:11:20,357][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:11:20,856][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:11:21,354][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:11:21,852][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:11:22,350][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:11:22,848][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:11:23,346][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:11:23,846][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:11:24,345][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:11:24,845][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:11:25,344][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:11:25,843][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:11:26,343][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:11:26,842][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:11:27,344][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:11:27,843][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:11:28,341][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:11:28,841][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:11:29,338][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:11:29,836][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:11:30,334][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:11:30,831][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:11:31,330][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:11:31,829][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:11:32,328][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:11:32,828][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:11:33,328][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:11:33,827][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:11:34,326][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:11:34,826][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:11:35,325][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:11:35,823][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:11:36,322][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:11:36,821][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:11:37,319][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:11:37,816][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:11:38,314][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:11:38,812][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:11:39,311][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:11:39,809][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:11:40,307][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:11:40,805][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:11:41,303][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:11:41,801][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:11:42,299][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:11:42,796][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:11:43,294][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:11:43,792][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:11:44,292][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:11:44,791][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:11:45,290][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:11:45,792][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:11:46,292][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:11:46,791][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:11:47,290][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:11:47,788][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:11:48,288][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:11:48,787][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:11:49,286][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:11:49,784][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:11:50,282][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:11:50,780][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:11:51,278][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:11:51,777][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:11:52,275][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:11:52,774][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:11:53,271][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:11:53,770][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:11:54,268][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:11:54,767][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:11:55,266][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:11:55,765][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:11:56,264][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:11:56,763][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:11:57,263][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:11:57,762][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:11:58,262][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:11:58,762][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:11:59,262][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:11:59,763][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:12:00,262][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:12:00,760][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:12:01,259][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:12:01,756][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:12:02,254][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:12:02,752][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:12:03,250][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:12:03,750][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:12:04,249][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:12:04,748][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:12:05,248][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:12:05,747][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:12:06,246][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:12:06,745][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:12:07,245][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:12:07,745][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:12:08,243][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:12:08,741][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:12:09,240][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:12:09,738][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:12:10,236][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:12:10,733][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:12:11,232][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:12:11,731][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:12:12,229][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:12:12,727][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:12:13,225][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:12:13,723][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:12:14,221][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:12:14,718][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:12:15,216][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:12:15,714][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:12:16,213][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:12:16,711][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:12:17,210][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:12:17,709][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:12:18,210][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:12:18,710][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:12:19,208][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:12:19,856][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 00:12:20,594][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:12:20,597][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:12:20,598][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:12:21,373][__main__][INFO] - Iteration 167 took 1m 13s (8.21% Gen, 90.74% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 53m 44s. Estimated total time: 61h 31m 23s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 2s, 500 more iterations: 10h 15m 13s. [2026-03-26 00:12:21,376][__main__][INFO] - Starting iteration 167. [2026-03-26 00:12:21,774][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:12:21,775][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:12:28,818][__main__][INFO] - Number of regex retries in iteration 167: 0 [2026-03-26 00:12:28,819][__main__][INFO] - agents played in iteration 167 are Alice, Bob [2026-03-26 00:12:29,692][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:12:29,736][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:12:29,737][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:12:29,737][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:12:30,348][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:12:30,803][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:12:31,305][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:12:31,805][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:12:32,304][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:12:32,803][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:12:33,302][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:12:33,802][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:12:34,300][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:12:34,799][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:12:35,297][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:12:35,796][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:12:36,294][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:12:36,793][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:12:37,292][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:12:37,791][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:12:38,290][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:12:38,787][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:12:39,284][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:12:39,782][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:12:40,280][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:12:40,778][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:12:41,275][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:12:41,772][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:12:42,270][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:12:42,767][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:12:43,264][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:12:43,762][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:12:44,259][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:12:44,756][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:12:45,254][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:12:45,751][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:12:46,250][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:12:46,747][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:12:47,245][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:12:47,742][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:12:48,241][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:12:48,740][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:12:49,238][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:12:49,736][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:12:50,234][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:12:50,733][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:12:51,230][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:12:51,729][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:12:52,228][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:12:52,725][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:12:53,224][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:12:53,723][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:12:54,222][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:12:54,721][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:12:55,220][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:12:55,718][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:12:56,216][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:12:56,714][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:12:57,212][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:12:57,711][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:12:58,210][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:12:58,708][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:12:59,206][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:12:59,702][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:13:00,200][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:13:00,697][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:13:01,194][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:13:01,691][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:13:02,188][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:13:02,686][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:13:03,185][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:13:03,683][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:13:04,183][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:13:04,681][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:13:05,179][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:13:05,677][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:13:06,175][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:13:06,674][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:13:07,172][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:13:07,672][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:13:08,171][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:13:08,670][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:13:09,169][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:13:09,668][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:13:10,166][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:13:10,665][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:13:11,163][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:13:11,660][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:13:12,158][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:13:12,655][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:13:13,152][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:13:13,649][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:13:14,147][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:13:14,645][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:13:15,145][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:13:15,646][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:13:16,145][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:13:16,643][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:13:17,140][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:13:17,637][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:13:18,135][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:13:18,632][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:13:19,129][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:13:19,626][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:13:20,123][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:13:20,620][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:13:21,117][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:13:21,613][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:13:22,117][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:13:22,616][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:13:23,116][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:13:23,614][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:13:24,113][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:13:24,613][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:13:25,112][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:13:25,612][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:13:26,112][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:13:26,611][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:13:27,110][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:13:27,609][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:13:28,109][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:13:28,614][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:13:29,115][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:13:29,616][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:13:30,116][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:13:30,616][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:13:31,115][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:13:31,613][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:13:32,112][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:13:32,610][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:13:33,108][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:13:33,607][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:13:34,106][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 00:13:34,779][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 00:13:35,527][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:13:35,529][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:13:35,531][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:13:36,199][__main__][INFO] - Iteration 168 took 1m 14s (9.46% Gen, 89.64% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 58h 22m 22s. Estimated total time: 62h 1m 16s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 2s, 500 more iterations: 10h 20m 12s. [2026-03-26 00:13:36,202][__main__][INFO] - Starting iteration 168. [2026-03-26 00:13:36,600][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:13:36,601][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:13:37,196][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:13:37,201][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:13:43,284][__main__][INFO] - Number of regex retries in iteration 168: 2 [2026-03-26 00:13:43,284][__main__][INFO] - agents played in iteration 168 are Alice, Bob [2026-03-26 00:13:44,162][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:13:44,206][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:13:44,207][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:13:44,207][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:13:44,826][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:13:45,286][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:13:45,792][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:13:46,293][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:13:46,794][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:13:47,294][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:13:47,795][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:13:48,298][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:13:48,799][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:13:49,298][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:13:49,799][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:13:50,298][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:13:50,798][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:13:51,298][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:13:51,798][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:13:52,298][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:13:52,797][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:13:53,295][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:13:53,793][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:13:54,292][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:13:54,790][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:13:55,289][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:13:55,788][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:13:56,286][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:13:56,783][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:13:57,283][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:13:57,782][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:13:58,281][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:13:58,778][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:13:59,276][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:13:59,774][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:14:00,272][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:14:00,770][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:14:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:14:01,768][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:14:02,268][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:14:02,767][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:14:03,265][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:14:03,763][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:14:04,260][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:14:04,758][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:14:05,256][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:14:05,754][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:14:06,253][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:14:06,751][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:14:07,249][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:14:07,748][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:14:08,247][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:14:08,746][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:14:09,244][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:14:09,745][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:14:10,244][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:14:10,744][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:14:11,243][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:14:11,741][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:14:12,239][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:14:12,738][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:14:13,237][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:14:13,736][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:14:14,234][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:14:14,733][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:14:15,231][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:14:15,730][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:14:16,228][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:14:16,726][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:14:17,227][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:14:17,727][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:14:18,229][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:14:18,731][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:14:19,231][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:14:19,731][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:14:20,231][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:14:20,732][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:14:21,233][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:14:21,733][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:14:22,232][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:14:22,731][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:14:23,230][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:14:23,729][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:14:24,228][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:14:24,727][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:14:25,226][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:14:25,725][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:14:26,222][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:14:26,720][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:14:27,219][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:14:27,718][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:14:28,215][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:14:28,714][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:14:29,211][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:14:29,709][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:14:30,207][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:14:30,704][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:14:31,201][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:14:31,699][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:14:32,195][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:14:32,692][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:14:33,191][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:14:33,689][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:14:34,187][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:14:34,685][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:14:35,183][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:14:35,680][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:14:36,178][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:14:36,676][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:14:37,174][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:14:37,672][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:14:38,171][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:14:38,669][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:14:39,167][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:14:39,665][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:14:40,162][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:14:40,660][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:14:41,158][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:14:41,657][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:14:42,155][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:14:42,653][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:14:43,153][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:14:43,652][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:14:44,151][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:14:44,652][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:14:45,154][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:14:45,657][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:14:46,159][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:14:46,660][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:14:47,161][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:14:47,660][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:14:48,161][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:14:48,663][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 00:14:49,319][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.61%, ΔTime: 00:01:04 [2026-03-26 00:14:50,059][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:14:50,061][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:14:50,063][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:14:50,787][__main__][INFO] - Iteration 169 took 1m 14s (9.01% Gen, 90.01% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 9m 12s. Estimated total time: 61h 49m 20s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 38s, 500 more iterations: 10h 18m 13s. [2026-03-26 00:14:50,789][__main__][INFO] - Starting iteration 169. [2026-03-26 00:14:51,186][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:14:51,187][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:14:57,837][__main__][INFO] - Number of regex retries in iteration 169: 0 [2026-03-26 00:14:57,838][__main__][INFO] - agents played in iteration 169 are Alice, Bob [2026-03-26 00:14:58,721][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:14:58,766][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:14:58,767][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:14:58,767][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:14:59,385][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:14:59,839][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:15:00,343][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:15:00,844][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:15:01,344][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:15:01,843][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:15:02,342][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:15:02,840][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:15:03,338][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:15:03,835][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:15:04,332][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:15:04,830][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:15:05,329][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:15:05,826][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:15:06,325][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:15:06,823][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:15:07,321][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:15:07,820][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:15:08,317][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:15:08,815][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:15:09,313][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:15:09,811][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:15:10,309][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:15:10,808][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:15:11,305][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:15:11,803][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:15:12,301][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:15:12,798][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:15:13,295][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:15:13,792][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:15:14,290][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:15:14,789][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:15:15,287][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:15:15,785][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:15:16,283][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:15:16,781][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:15:17,279][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:15:17,777][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:15:18,276][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:15:18,777][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:15:19,276][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:15:19,776][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:15:20,276][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:15:20,776][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:15:21,276][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:15:21,776][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:15:22,276][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:15:22,776][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:15:23,276][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:15:23,774][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:15:24,272][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:15:24,770][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:15:25,270][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:15:25,769][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:15:26,267][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:15:26,764][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:15:27,262][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:15:27,760][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:15:28,258][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:15:28,756][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:15:29,254][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:15:29,752][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:15:30,251][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:15:30,749][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:15:31,248][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:15:31,745][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:15:32,243][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:15:32,741][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:15:33,238][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:15:33,736][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:15:34,233][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:15:34,731][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:15:35,229][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:15:35,726][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:15:36,225][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:15:36,725][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:15:37,222][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:15:37,719][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:15:38,217][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:15:38,715][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:15:39,212][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:15:39,710][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:15:40,208][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:15:40,706][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:15:41,204][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:15:41,702][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:15:42,200][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:15:42,698][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:15:43,196][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:15:43,694][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:15:44,192][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:15:44,690][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:15:45,188][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:15:45,686][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:15:46,183][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:15:46,680][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:15:47,178][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:15:47,676][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:15:48,177][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:15:48,677][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:15:49,176][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:15:49,677][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:15:50,175][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:15:50,674][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:15:51,173][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:15:51,674][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:15:52,174][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:15:52,673][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:15:53,171][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:15:53,670][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:15:54,170][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:15:54,670][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:15:55,170][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:15:55,668][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:15:56,166][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:15:56,664][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:15:57,161][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:15:57,659][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:15:58,156][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:15:58,654][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:15:59,152][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:15:59,650][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:16:00,148][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:16:00,646][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:16:01,144][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:16:01,641][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:16:02,138][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:16:02,635][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:16:03,133][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 00:16:03,796][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:16:04,541][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:16:04,544][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:16:04,545][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:16:05,379][__main__][INFO] - Iteration 170 took 1m 14s (8.96% Gen, 89.91% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 8m 16s. Estimated total time: 61h 49m 39s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 16s. [2026-03-26 00:16:05,382][__main__][INFO] - Starting iteration 170. [2026-03-26 00:16:05,782][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:16:05,783][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:16:11,899][mllm.models.large_language_model_local][WARNING] - Response Given the current situation, it's beneficial to align our proposals to match Alice's to maximize our points, especially since our per-item values for books and balls don't match hers. We should propose the same distribution she does to avoid any proportional allocation. Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:16:12,998][__main__][INFO] - Number of regex retries in iteration 170: 1 [2026-03-26 00:16:12,999][__main__][INFO] - agents played in iteration 170 are Alice, Bob [2026-03-26 00:16:13,870][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:16:13,914][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:16:13,915][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:16:13,915][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:16:14,515][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:16:14,970][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:16:15,475][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:16:15,976][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:16:16,476][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:16:16,974][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:16:17,472][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:16:17,970][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:16:18,472][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:16:18,970][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:16:19,468][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:16:19,966][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:16:20,465][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:16:20,962][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:16:21,461][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:16:21,958][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:16:22,455][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:16:22,953][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:16:23,450][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:16:23,948][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:16:24,445][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:16:24,943][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:16:25,440][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:16:25,937][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:16:26,434][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:16:26,932][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:16:27,431][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:16:27,929][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:16:28,429][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:16:28,927][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:16:29,425][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:16:29,924][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:16:30,422][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:16:30,919][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:16:31,417][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:16:31,915][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:16:32,413][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:16:32,912][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:16:33,410][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:16:33,908][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:16:34,405][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:16:34,902][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:16:35,398][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:16:35,895][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:16:36,393][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:16:36,891][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:16:37,389][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:16:37,886][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:16:38,384][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:16:38,881][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:16:39,378][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:16:39,875][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:16:40,372][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:16:40,870][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:16:41,369][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:16:41,866][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:16:42,364][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:16:42,862][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:16:43,359][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:16:43,857][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:16:44,353][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:16:44,851][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:16:45,348][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:16:45,847][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:16:46,344][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:16:46,843][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:16:47,341][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:16:47,838][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:16:48,338][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:16:48,836][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:16:49,334][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:16:49,832][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:16:50,331][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:16:50,830][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:16:51,327][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:16:51,825][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:16:52,322][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:16:52,819][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:16:53,317][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:16:53,814][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:16:54,312][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:16:54,809][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:16:55,307][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:16:55,805][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:16:56,303][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:16:56,800][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:16:57,296][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:16:57,793][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:16:58,291][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:16:58,790][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:16:59,290][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:16:59,788][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:17:00,287][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:17:00,786][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:17:01,284][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:17:01,782][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:17:02,280][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:17:02,778][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:17:03,275][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:17:03,773][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:17:04,273][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:17:04,771][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:17:05,269][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:17:05,767][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:17:06,266][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:17:06,765][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:17:07,263][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:17:07,762][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:17:08,259][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:17:08,756][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:17:09,253][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:17:09,751][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:17:10,248][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:17:10,746][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:17:11,244][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:17:11,742][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:17:12,239][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:17:12,736][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:17:13,236][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:17:13,734][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:17:14,232][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:17:14,731][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:17:15,229][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:17:15,728][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:17:16,228][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:17:16,726][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:17:17,226][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:17:17,724][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:17:18,223][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 00:17:18,862][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 00:17:19,597][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:17:19,599][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:17:19,600][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:17:20,256][__main__][INFO] - Iteration 171 took 1m 14s (9.69% Gen, 89.43% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 58h 21m 3s. Estimated total time: 62h 3m 41s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 7s, 500 more iterations: 10h 20m 36s. [2026-03-26 00:17:20,258][__main__][INFO] - Starting iteration 171. [2026-03-26 00:17:20,656][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:17:20,657][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:17:23,200][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:17:24,297][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:17:26,685][__main__][INFO] - Number of regex retries in iteration 171: 2 [2026-03-26 00:17:26,686][__main__][INFO] - agents played in iteration 171 are Alice, Bob [2026-03-26 00:17:27,566][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:17:27,896][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:17:27,897][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:17:27,897][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:17:28,502][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:17:28,960][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:17:29,466][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:17:29,967][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:17:30,468][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:17:30,969][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:17:31,470][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:17:31,972][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:17:32,473][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:17:32,973][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:17:33,474][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:17:33,973][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:17:34,471][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:17:34,971][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:17:35,470][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:17:35,968][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:17:36,467][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:17:36,967][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:17:37,466][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:17:37,966][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:17:38,466][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:17:38,966][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:17:39,466][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:17:39,966][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:17:40,466][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:17:40,964][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:17:41,461][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:17:41,958][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:17:42,455][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:17:42,951][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:17:43,449][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:17:43,946][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:17:44,444][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:17:44,942][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:17:45,442][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:17:45,943][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:17:46,443][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:17:46,940][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:17:47,439][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:17:47,937][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:17:48,436][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:17:48,935][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:17:49,432][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:17:49,930][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:17:50,428][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:17:50,928][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:17:51,427][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:17:51,928][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:17:52,426][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:17:52,925][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:17:53,422][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:17:53,920][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:17:54,419][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:17:54,916][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:17:55,414][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:17:55,912][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:17:56,411][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:17:56,909][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:17:57,407][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:17:57,905][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:17:58,404][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:17:58,902][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:17:59,399][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:17:59,896][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:18:00,395][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:18:00,894][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:18:01,394][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:18:01,893][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:18:02,393][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:18:02,891][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:18:03,391][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:18:03,890][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:18:04,390][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:18:04,888][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:18:05,386][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:18:05,885][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:18:06,383][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:18:06,880][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:18:07,378][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:18:07,875][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:18:08,374][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:18:08,874][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:18:09,373][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:18:09,872][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:18:10,371][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:18:10,870][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:18:11,369][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:18:11,867][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:18:12,367][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:18:12,866][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:18:13,363][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:18:13,860][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:18:14,357][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:18:14,854][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:18:15,351][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:18:15,848][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:18:16,346][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:18:16,846][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:18:17,345][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:18:17,843][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:18:18,344][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:18:18,842][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:18:19,340][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:18:19,838][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:18:20,335][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:18:20,833][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:18:21,331][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:18:21,829][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:18:22,327][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:18:22,824][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:18:23,322][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:18:23,819][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:18:24,316][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:18:24,814][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:18:25,311][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:18:25,810][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:18:26,308][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:18:26,806][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:18:27,304][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:18:27,802][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:18:28,300][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:18:28,798][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:18:29,296][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:18:29,794][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:18:30,293][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:18:30,791][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:18:31,289][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:18:31,788][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:18:32,286][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21760 tokens. [2026-03-26 00:18:32,925][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:18:33,762][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:18:33,764][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:18:33,766][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:18:34,515][__main__][INFO] - Iteration 172 took 1m 13s (8.16% Gen, 90.82% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 57h 49m 3s. Estimated total time: 61h 32m 56s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 5s, 500 more iterations: 10h 15m 29s. [2026-03-26 00:18:34,517][__main__][INFO] - Starting iteration 172. [2026-03-26 00:18:34,914][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:18:34,915][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:18:41,658][__main__][INFO] - Number of regex retries in iteration 172: 0 [2026-03-26 00:18:41,659][__main__][INFO] - agents played in iteration 172 are Alice, Bob [2026-03-26 00:18:42,533][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:18:42,577][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:18:42,578][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:18:42,579][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:18:43,189][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:18:43,643][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:18:44,147][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:18:44,647][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:18:45,147][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:18:45,646][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:18:46,145][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:18:46,643][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:18:47,142][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:18:47,640][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:18:48,138][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:18:48,637][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:18:49,135][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:18:49,632][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:18:50,130][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:18:50,628][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:18:51,125][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:18:51,626][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:18:52,128][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:18:52,627][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:18:53,128][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:18:53,628][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:18:54,127][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:18:54,627][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:18:55,127][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:18:55,626][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:18:56,124][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:18:56,622][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:18:57,120][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:18:57,617][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:18:58,115][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:18:58,613][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:18:59,111][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:18:59,610][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:19:00,108][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:19:00,606][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:19:01,105][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:19:01,604][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:19:02,102][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:19:02,601][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:19:03,099][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:19:03,597][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:19:04,095][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:19:04,593][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:19:05,092][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:19:05,591][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:19:06,089][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:19:06,586][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:19:07,085][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:19:07,582][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:19:08,079][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:19:08,576][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:19:09,074][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:19:09,572][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:19:10,070][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:19:10,568][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:19:11,066][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:19:11,566][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:19:12,066][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:19:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:19:13,067][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:19:13,566][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:19:14,066][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:19:14,565][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:19:15,065][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:19:15,568][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:19:16,069][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:19:16,569][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:19:17,069][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:19:17,567][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:19:18,066][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:19:18,568][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:19:19,067][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:19:19,567][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:19:20,066][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:19:20,563][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:19:21,060][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:19:21,557][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:19:22,055][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:19:22,552][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:19:23,049][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:19:23,548][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:19:24,048][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:19:24,547][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:19:25,048][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:19:25,547][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:19:26,047][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:19:26,546][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:19:27,046][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:19:27,545][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:19:28,045][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:19:28,542][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:19:29,040][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:19:29,537][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:19:30,035][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:19:30,532][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:19:31,030][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:19:31,529][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:19:32,028][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:19:32,527][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:19:33,026][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:19:33,525][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:19:34,024][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:19:34,523][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:19:35,023][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:19:35,521][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:19:36,019][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:19:36,517][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:19:37,015][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:19:37,514][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:19:38,012][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:19:38,510][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:19:39,009][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:19:39,507][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:19:40,005][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:19:40,503][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:19:41,000][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:19:41,497][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:19:41,994][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:19:42,492][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:19:42,989][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:19:43,489][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:19:43,989][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:19:44,488][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:19:44,988][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:19:45,488][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:19:45,987][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:19:46,487][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:19:46,985][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 00:19:47,614][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 00:19:48,353][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:19:48,355][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:19:48,357][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:19:49,140][__main__][INFO] - Iteration 173 took 1m 14s (9.09% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 6m 12s. Estimated total time: 61h 51m 19s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 42s, 500 more iterations: 10h 18m 33s. [2026-03-26 00:19:49,143][__main__][INFO] - Starting iteration 173. [2026-03-26 00:19:49,545][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:19:49,546][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:19:53,521][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:19:56,489][__main__][INFO] - Number of regex retries in iteration 173: 1 [2026-03-26 00:19:56,490][__main__][INFO] - agents played in iteration 173 are Alice, Bob [2026-03-26 00:19:57,365][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:19:57,409][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:19:57,410][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:19:57,410][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:19:58,019][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:19:58,476][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:19:58,981][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:19:59,481][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:19:59,980][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:20:00,479][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:20:00,978][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:20:01,477][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:20:01,975][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:20:02,474][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:20:02,973][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:20:03,471][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:20:03,969][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:20:04,467][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:20:04,966][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:20:05,464][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:20:05,963][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:20:06,461][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:20:06,960][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:20:07,458][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:20:07,957][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:20:08,455][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:20:08,954][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:20:09,452][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:20:09,951][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:20:10,449][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:20:10,947][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:20:11,444][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:20:11,941][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:20:12,438][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:20:12,934][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:20:13,431][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:20:13,929][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:20:14,427][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:20:14,924][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:20:15,421][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:20:15,919][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:20:16,416][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:20:16,914][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:20:17,411][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:20:17,909][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:20:18,409][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:20:18,908][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:20:19,405][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:20:19,904][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:20:20,403][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:20:20,900][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:20:21,398][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:20:21,896][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:20:22,394][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:20:22,892][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:20:23,390][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:20:23,888][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:20:24,386][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:20:24,885][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:20:25,384][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:20:25,883][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:20:26,379][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:20:26,876][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:20:27,374][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:20:27,871][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:20:28,368][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:20:28,866][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:20:29,363][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:20:29,859][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:20:30,358][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:20:30,856][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:20:31,353][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:20:31,852][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:20:32,351][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:20:32,850][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:20:33,348][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:20:33,847][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:20:34,346][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:20:34,844][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:20:35,342][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:20:35,840][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:20:36,337][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:20:36,834][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:20:37,333][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:20:37,831][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:20:38,329][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:20:38,827][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:20:39,325][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:20:39,823][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:20:40,321][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:20:40,820][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:20:41,318][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:20:41,816][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:20:42,315][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:20:42,812][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:20:43,310][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:20:43,807][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:20:44,305][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:20:44,802][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:20:45,300][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:20:45,800][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:20:46,298][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:20:46,795][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:20:47,291][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:20:47,788][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:20:48,288][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:20:48,785][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:20:49,282][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:20:49,779][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:20:50,276][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:20:50,775][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:20:51,273][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:20:51,771][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:20:52,269][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:20:52,767][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:20:53,265][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:20:53,762][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:20:54,260][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:20:54,758][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:20:55,256][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:20:55,755][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:20:56,253][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:20:56,751][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:20:57,249][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:20:57,748][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:20:58,246][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:20:58,744][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:20:59,240][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:20:59,737][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:21:00,235][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:21:00,732][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:21:01,229][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:21:01,726][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 00:21:02,356][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 00:21:03,088][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:21:03,090][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:21:03,093][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:21:03,810][__main__][INFO] - Iteration 174 took 1m 14s (9.35% Gen, 89.68% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 6m 55s. Estimated total time: 61h 53m 16s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 46s, 500 more iterations: 10h 18m 52s. [2026-03-26 00:21:03,812][__main__][INFO] - Starting iteration 174. [2026-03-26 00:21:04,210][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:21:04,211][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:21:06,422][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:21:11,180][__main__][INFO] - Number of regex retries in iteration 174: 1 [2026-03-26 00:21:11,180][__main__][INFO] - agents played in iteration 174 are Alice, Bob [2026-03-26 00:21:12,055][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:21:12,099][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:21:12,100][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:21:12,100][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:21:12,707][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:21:13,160][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:21:13,664][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:21:14,161][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:21:14,658][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:21:15,155][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:21:15,652][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:21:16,149][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:21:16,645][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:21:17,142][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:21:17,639][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:21:18,137][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:21:18,635][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:21:19,133][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:21:19,630][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:21:20,128][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:21:20,626][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:21:21,126][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:21:21,626][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:21:22,125][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:21:22,624][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:21:23,123][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:21:23,622][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:21:24,119][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:21:24,617][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:21:25,117][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:21:25,615][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:21:26,114][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:21:26,613][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:21:27,112][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:21:27,610][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:21:28,110][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:21:28,608][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:21:29,106][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:21:29,603][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:21:30,101][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:21:30,598][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:21:31,097][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:21:31,595][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:21:32,093][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:21:32,591][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:21:33,090][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:21:33,589][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:21:34,088][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:21:34,587][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:21:35,087][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:21:35,586][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:21:36,085][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:21:36,584][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:21:37,081][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:21:37,576][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:21:38,074][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:21:38,571][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:21:39,069][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:21:39,567][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:21:40,066][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:21:40,563][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:21:41,061][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:21:41,559][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:21:42,057][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:21:42,555][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:21:43,053][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:21:43,552][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:21:44,050][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:21:44,548][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:21:45,046][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:21:45,544][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:21:46,041][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:21:46,539][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:21:47,036][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:21:47,533][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:21:48,032][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:21:48,532][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:21:49,029][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:21:49,527][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:21:50,025][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:21:50,523][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:21:51,022][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:21:51,520][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:21:52,018][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:21:52,515][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:21:53,013][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:21:53,512][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:21:54,011][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:21:54,510][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:21:55,009][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:21:55,508][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:21:56,008][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:21:56,508][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:21:57,007][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:21:57,506][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:21:58,005][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:21:58,504][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:21:59,002][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:21:59,500][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:21:59,998][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:22:00,497][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:22:00,995][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:22:01,493][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:22:01,990][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:22:02,488][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:22:02,986][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:22:03,484][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:22:03,981][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:22:04,478][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:22:04,976][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:22:05,476][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:22:05,975][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:22:06,474][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:22:06,974][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:22:07,474][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:22:07,973][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:22:08,472][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:22:08,970][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:22:09,467][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:22:09,965][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:22:10,462][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:22:10,960][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:22:11,457][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:22:11,954][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:22:12,451][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:22:12,949][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:22:13,449][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:22:13,947][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:22:14,446][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:22:14,945][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:22:15,443][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:22:15,943][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:22:16,441][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21733 tokens. [2026-03-26 00:22:17,074][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 00:22:17,806][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:22:17,809][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:22:17,810][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:22:18,466][__main__][INFO] - Iteration 175 took 1m 14s (9.39% Gen, 89.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 58h 5m 12s. Estimated total time: 61h 52m 49s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 45s, 500 more iterations: 10h 18m 48s. [2026-03-26 00:22:18,468][__main__][INFO] - Starting iteration 175. [2026-03-26 00:22:18,866][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:22:18,867][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:22:25,435][__main__][INFO] - Number of regex retries in iteration 175: 0 [2026-03-26 00:22:25,436][__main__][INFO] - agents played in iteration 175 are Alice, Bob [2026-03-26 00:22:26,317][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:22:26,361][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:22:26,362][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:22:26,362][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:22:26,964][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:22:27,418][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:22:27,920][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:22:28,418][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:22:28,917][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:22:29,416][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:22:29,915][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:22:30,413][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:22:30,912][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:22:31,410][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:22:31,908][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:22:32,406][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:22:32,907][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:22:33,407][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:22:33,905][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:22:34,403][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:22:34,902][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:22:35,399][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:22:35,896][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:22:36,394][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:22:36,892][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:22:37,390][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:22:37,888][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:22:38,386][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:22:38,884][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:22:39,384][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:22:39,883][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:22:40,381][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:22:40,879][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:22:41,377][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:22:41,876][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:22:42,375][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:22:42,873][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:22:43,371][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:22:43,868][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:22:44,365][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:22:44,862][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:22:45,359][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:22:45,856][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:22:46,353][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:22:46,851][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:22:47,351][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:22:47,850][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:22:48,352][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:22:48,852][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:22:49,351][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:22:49,850][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:22:50,350][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:22:50,849][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:22:51,346][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:22:51,844][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:22:52,342][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:22:52,839][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:22:53,337][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:22:53,835][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:22:54,334][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:22:54,832][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:22:55,330][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:22:55,827][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:22:56,326][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:22:56,822][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:22:57,320][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:22:57,817][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:22:58,315][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:22:58,813][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:22:59,311][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:22:59,810][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:23:00,308][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:23:00,807][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:23:01,305][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:23:01,803][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:23:02,300][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:23:02,799][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:23:03,297][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:23:03,796][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:23:04,294][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:23:04,793][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:23:05,291][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:23:05,790][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:23:06,288][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:23:06,787][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:23:07,286][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:23:07,784][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:23:08,283][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:23:08,781][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:23:09,280][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:23:09,778][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:23:10,276][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:23:10,773][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:23:11,272][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:23:11,770][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:23:12,269][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:23:12,768][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:23:13,267][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:23:13,765][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:23:14,263][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:23:14,761][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:23:15,259][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:23:15,756][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:23:16,254][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:23:16,752][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:23:17,250][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:23:17,747][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:23:18,247][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:23:18,747][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:23:19,247][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:23:19,748][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:23:20,248][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:23:20,746][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:23:21,245][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:23:21,744][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:23:22,243][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:23:22,741][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:23:23,239][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:23:23,736][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:23:24,234][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:23:24,732][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:23:25,229][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:23:25,728][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:23:26,226][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:23:26,724][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:23:27,222][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:23:27,719][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:23:28,217][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:23:28,715][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:23:29,214][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:23:29,713][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:23:30,211][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:23:30,709][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 00:23:31,335][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:04 [2026-03-26 00:23:32,064][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:23:32,066][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:23:32,067][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:23:32,902][__main__][INFO] - Iteration 176 took 1m 14s (8.87% Gen, 90.00% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 52m 58s. Estimated total time: 61h 41m 49s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 23s, 500 more iterations: 10h 16m 58s. [2026-03-26 00:23:32,905][__main__][INFO] - Starting iteration 176. [2026-03-26 00:23:33,304][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:23:33,305][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:23:34,942][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:23:37,017][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:23:39,824][__main__][INFO] - Number of regex retries in iteration 176: 2 [2026-03-26 00:23:39,824][__main__][INFO] - agents played in iteration 176 are Alice, Bob [2026-03-26 00:23:40,700][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:23:40,744][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:23:40,745][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:23:40,745][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:23:41,350][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:23:41,805][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:23:42,308][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:23:42,806][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:23:43,304][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:23:43,802][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:23:44,299][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:23:44,798][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:23:45,297][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:23:45,800][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:23:46,300][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:23:46,800][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:23:47,298][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:23:47,797][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:23:48,300][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:23:48,799][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:23:49,300][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:23:49,797][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:23:50,295][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:23:50,794][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:23:51,292][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:23:51,791][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:23:52,289][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:23:52,788][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:23:53,286][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:23:53,784][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:23:54,282][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:23:54,779][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:23:55,277][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:23:55,774][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:23:56,271][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:23:56,770][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:23:57,269][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:23:57,768][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:23:58,266][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:23:58,766][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:23:59,264][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:23:59,762][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:24:00,259][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:24:00,757][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:24:01,256][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:24:01,754][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:24:02,252][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:24:02,750][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:24:03,249][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:24:03,747][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:24:04,246][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:24:04,744][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:24:05,241][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:24:05,738][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:24:06,236][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:24:06,734][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:24:07,231][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:24:07,728][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:24:08,224][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:24:08,721][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:24:09,218][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:24:09,717][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:24:10,216][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:24:10,716][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:24:11,215][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:24:11,714][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:24:12,213][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:24:12,712][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:24:13,212][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:24:13,710][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:24:14,209][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:24:14,707][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:24:15,207][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:24:15,704][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:24:16,202][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:24:16,700][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:24:17,198][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:24:17,697][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:24:18,198][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:24:18,697][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:24:19,196][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:24:19,696][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:24:20,195][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:24:20,694][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:24:21,194][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:24:21,692][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:24:22,191][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:24:22,689][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:24:23,188][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:24:23,686][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:24:24,184][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:24:24,681][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:24:25,180][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:24:25,677][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:24:26,176][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:24:26,673][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:24:27,171][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:24:27,668][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:24:28,165][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:24:28,662][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:24:29,159][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:24:29,657][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:24:30,156][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:24:30,655][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:24:31,154][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:24:31,652][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:24:32,150][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:24:32,648][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:24:33,147][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:24:33,644][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:24:34,142][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:24:34,639][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:24:35,137][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:24:35,635][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:24:36,132][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:24:36,630][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:24:37,129][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:24:37,625][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:24:38,122][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:24:38,619][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:24:39,116][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:24:39,613][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:24:40,111][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:24:40,608][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:24:41,107][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:24:41,606][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:24:42,105][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:24:42,603][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:24:43,102][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:24:43,602][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:24:44,101][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:24:44,600][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:24:45,099][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 00:24:45,728][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:24:46,457][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:24:46,460][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:24:46,461][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:24:47,107][__main__][INFO] - Iteration 177 took 1m 13s (8.83% Gen, 90.29% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 40m 5s. Estimated total time: 61h 30m 10s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 0s, 500 more iterations: 10h 15m 1s. [2026-03-26 00:24:47,109][__main__][INFO] - Starting iteration 177. [2026-03-26 00:24:47,507][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:24:47,508][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:24:54,074][__main__][INFO] - Number of regex retries in iteration 177: 0 [2026-03-26 00:24:54,075][__main__][INFO] - agents played in iteration 177 are Alice, Bob [2026-03-26 00:24:54,949][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:24:54,993][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:24:54,994][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:24:54,994][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:24:55,600][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:24:56,057][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:24:56,563][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:24:57,064][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:24:57,564][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:24:58,064][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:24:58,566][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:24:59,071][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:24:59,573][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:25:00,071][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:25:00,569][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:25:01,068][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:25:01,566][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:25:02,065][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:25:02,562][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:25:03,059][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:25:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:25:04,056][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:25:04,554][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:25:05,052][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:25:05,550][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:25:06,051][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:25:06,549][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:25:07,047][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:25:07,545][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:25:08,044][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:25:08,543][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:25:09,042][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:25:09,540][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:25:10,039][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:25:10,536][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:25:11,036][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:25:11,535][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:25:12,033][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:25:12,530][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:25:13,028][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:25:13,525][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:25:14,021][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:25:14,519][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:25:15,017][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:25:15,517][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:25:16,017][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:25:16,516][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:25:17,014][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:25:17,511][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:25:18,008][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:25:18,508][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:25:19,005][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:25:19,503][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:25:20,002][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:25:20,501][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:25:21,000][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:25:21,498][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:25:21,997][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:25:22,496][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:25:22,995][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:25:23,494][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:25:23,993][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:25:24,491][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:25:24,991][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:25:25,490][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:25:25,988][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:25:26,487][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:25:26,985][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:25:27,483][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:25:27,985][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:25:28,487][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:25:28,989][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:25:29,490][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:25:29,991][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:25:30,493][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:25:30,995][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:25:31,497][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:25:31,996][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:25:32,495][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:25:32,993][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:25:33,490][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:25:33,988][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:25:34,488][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:25:34,988][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:25:35,487][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:25:35,985][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:25:36,484][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:25:36,982][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:25:37,479][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:25:37,977][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:25:38,476][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:25:38,975][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:25:39,473][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:25:39,972][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:25:40,471][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:25:40,969][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:25:41,468][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:25:41,968][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:25:42,468][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:25:42,969][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:25:43,468][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:25:43,967][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:25:44,465][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:25:44,962][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:25:45,459][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:25:45,957][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:25:46,455][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:25:46,954][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:25:47,452][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:25:47,951][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:25:48,450][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:25:48,948][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:25:49,445][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:25:49,942][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:25:50,439][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:25:50,937][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:25:51,435][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:25:51,934][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:25:52,432][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:25:52,929][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:25:53,426][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:25:53,924][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:25:54,421][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:25:54,919][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:25:55,418][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:25:55,916][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:25:56,414][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:25:56,913][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:25:57,411][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:25:57,909][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:25:58,408][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:25:58,905][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:25:59,403][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:26:00,033][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 00:26:00,769][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:26:00,771][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:26:00,773][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:26:01,480][__main__][INFO] - Iteration 178 took 1m 13s (8.88% Gen, 90.16% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 47m 19s. Estimated total time: 61h 38m 39s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 17s, 500 more iterations: 10h 16m 26s. [2026-03-26 00:26:01,482][__main__][INFO] - Starting iteration 178. [2026-03-26 00:26:01,880][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:26:01,881][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:26:08,291][__main__][INFO] - Number of regex retries in iteration 178: 0 [2026-03-26 00:26:08,292][__main__][INFO] - agents played in iteration 178 are Alice, Bob [2026-03-26 00:26:09,168][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:26:09,490][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:26:09,491][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:26:09,491][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:26:10,105][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:26:10,561][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:26:11,064][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:26:11,562][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:26:12,061][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:26:12,560][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:26:13,059][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:26:13,557][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:26:14,057][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:26:14,556][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:26:15,055][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:26:15,553][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:26:16,053][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:26:16,552][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:26:17,051][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:26:17,550][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:26:18,049][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:26:18,548][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:26:19,045][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:26:19,542][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:26:20,039][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:26:20,536][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:26:21,034][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:26:21,532][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:26:22,030][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:26:22,530][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:26:23,028][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:26:23,526][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:26:24,024][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:26:24,523][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:26:25,023][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:26:25,522][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:26:26,021][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:26:26,520][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:26:27,019][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:26:27,518][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:26:28,016][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:26:28,515][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:26:29,015][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:26:29,514][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:26:30,014][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:26:30,516][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:26:31,016][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:26:31,516][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:26:32,017][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:26:32,517][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:26:33,017][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:26:33,518][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:26:34,018][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:26:34,516][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:26:35,014][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:26:35,513][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:26:36,011][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:26:36,511][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:26:37,010][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:26:37,509][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:26:38,008][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:26:38,506][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:26:39,003][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:26:39,502][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:26:39,999][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:26:40,497][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:26:40,995][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:26:41,494][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:26:41,992][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:26:42,491][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:26:42,989][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:26:43,488][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:26:43,986][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:26:44,485][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:26:44,984][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:26:45,485][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:26:45,987][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:26:46,486][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:26:46,984][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:26:47,482][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:26:47,981][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:26:48,483][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:26:48,982][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:26:49,482][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:26:49,980][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:26:50,477][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:26:50,975][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:26:51,472][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:26:51,971][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:26:52,470][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:26:52,968][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:26:53,468][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:26:53,966][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:26:54,464][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:26:54,963][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:26:55,461][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:26:55,960][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:26:56,458][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:26:56,958][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:26:57,458][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:26:57,957][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:26:58,456][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:26:58,954][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:26:59,453][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:26:59,952][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:27:00,452][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:27:00,951][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:27:01,451][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:27:01,950][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:27:02,449][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:27:02,949][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:27:03,449][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:27:03,950][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:27:04,450][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:27:04,950][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:27:05,449][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:27:05,948][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:27:06,447][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:27:06,945][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:27:07,443][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:27:07,942][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:27:08,440][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:27:08,938][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:27:09,437][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:27:09,935][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:27:10,433][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:27:10,932][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:27:11,430][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:27:11,929][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:27:12,429][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:27:12,926][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:27:13,425][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:27:13,923][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21745 tokens. [2026-03-26 00:27:14,564][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 00:27:15,301][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:27:15,304][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:27:15,305][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:27:16,169][__main__][INFO] - Iteration 179 took 1m 14s (8.63% Gen, 90.21% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 58h 1m 53s. Estimated total time: 61h 54m 27s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 48s, 500 more iterations: 10h 19m 4s. [2026-03-26 00:27:16,171][__main__][INFO] - Starting iteration 179. [2026-03-26 00:27:16,570][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:27:16,571][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:27:17,614][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 10 books, 30 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:27:23,003][__main__][INFO] - Number of regex retries in iteration 179: 1 [2026-03-26 00:27:23,004][__main__][INFO] - agents played in iteration 179 are Alice, Bob [2026-03-26 00:27:23,878][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:27:23,923][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:27:23,923][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:27:23,924][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:27:24,528][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:27:24,981][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:27:25,484][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:27:25,981][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:27:26,479][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:27:26,977][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:27:27,476][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:27:27,975][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:27:28,474][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:27:28,974][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:27:29,473][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:27:29,973][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:27:30,473][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:27:30,971][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:27:31,470][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:27:31,970][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:27:32,470][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:27:32,969][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:27:33,467][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:27:33,966][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:27:34,465][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:27:34,964][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:27:35,464][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:27:35,963][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:27:36,462][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:27:36,960][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:27:37,458][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:27:37,957][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:27:38,455][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:27:38,955][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:27:39,455][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:27:39,953][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:27:40,452][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:27:40,951][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:27:41,450][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:27:41,949][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:27:42,447][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:27:42,946][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:27:43,444][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:27:43,943][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:27:44,440][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:27:44,938][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:27:45,435][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:27:45,933][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:27:46,432][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:27:46,930][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:27:47,429][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:27:47,929][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:27:48,429][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:27:48,927][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:27:49,426][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:27:49,923][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:27:50,421][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:27:50,919][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:27:51,417][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:27:51,917][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:27:52,416][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:27:52,914][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:27:53,413][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:27:53,912][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:27:54,411][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:27:54,911][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:27:55,411][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:27:55,909][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:27:56,408][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:27:56,905][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:27:57,402][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:27:57,901][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:27:58,399][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:27:58,897][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:27:59,395][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:27:59,893][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:28:00,391][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:28:00,890][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:28:01,389][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:28:01,887][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:28:02,385][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:28:02,884][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:28:03,382][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:28:03,880][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:28:04,379][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:28:04,878][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:28:05,376][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:28:05,875][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:28:06,374][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:28:06,873][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:28:07,372][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:28:07,871][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:28:08,370][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:28:08,869][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:28:09,366][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:28:09,863][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:28:10,360][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:28:10,858][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:28:11,356][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:28:11,854][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:28:12,352][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:28:12,851][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:28:13,349][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:28:13,847][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:28:14,345][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:28:14,842][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:28:15,342][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:28:15,843][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:28:16,341][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:28:16,839][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:28:17,336][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:28:17,834][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:28:18,334][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:28:18,832][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:28:19,329][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:28:19,827][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:28:20,324][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:28:20,822][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:28:21,319][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:28:21,817][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:28:22,315][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:28:22,812][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:28:23,310][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:28:23,807][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:28:24,304][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:28:24,803][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:28:25,302][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:28:25,801][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:28:26,299][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:28:26,798][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:28:27,296][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:28:27,795][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:28:28,293][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 00:28:28,926][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 00:28:29,658][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:28:29,660][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:28:29,661][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:28:30,309][__main__][INFO] - Iteration 180 took 1m 13s (8.72% Gen, 90.40% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 33m 10s. Estimated total time: 61h 26m 58s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 53s, 500 more iterations: 10h 14m 29s. [2026-03-26 00:28:30,311][__main__][INFO] - Starting iteration 180. [2026-03-26 00:28:30,709][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:28:30,710][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:28:31,854][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:28:32,232][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:28:32,411][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:28:36,166][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:28:37,203][__main__][INFO] - Number of regex retries in iteration 180: 4 [2026-03-26 00:28:37,204][__main__][INFO] - agents played in iteration 180 are Alice, Bob [2026-03-26 00:28:38,076][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:28:38,120][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:28:38,121][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:28:38,122][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:28:38,730][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:28:39,185][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:28:39,688][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:28:40,188][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:28:40,690][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:28:41,191][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:28:41,691][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:28:42,191][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:28:42,690][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:28:43,188][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:28:43,687][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:28:44,185][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:28:44,685][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:28:45,184][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:28:45,683][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:28:46,181][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:28:46,681][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:28:47,181][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:28:47,681][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:28:48,184][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:28:48,685][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:28:49,186][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:28:49,686][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:28:50,185][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:28:50,684][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:28:51,184][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:28:51,682][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:28:52,181][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:28:52,680][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:28:53,179][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:28:53,678][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:28:54,176][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:28:54,675][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:28:55,176][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:28:55,676][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:28:56,176][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:28:56,676][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:28:57,175][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:28:57,673][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:28:58,173][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:28:58,672][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:28:59,171][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:28:59,670][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:29:00,168][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:29:00,667][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:29:01,165][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:29:01,663][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:29:02,161][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:29:02,659][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:29:03,158][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:29:03,656][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:29:04,157][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:29:04,656][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:29:05,154][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:29:05,652][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:29:06,152][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:29:06,650][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:29:07,151][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:29:07,649][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:29:08,147][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:29:08,645][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:29:09,142][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:29:09,639][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:29:10,137][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:29:10,635][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:29:11,134][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:29:11,633][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:29:12,132][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:29:12,632][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:29:13,131][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:29:13,629][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:29:14,128][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:29:14,627][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:29:15,126][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:29:15,624][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:29:16,123][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:29:16,622][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:29:17,121][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:29:17,619][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:29:18,118][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:29:18,619][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:29:19,119][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:29:19,618][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:29:20,119][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:29:20,618][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:29:21,118][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:29:21,617][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:29:22,117][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:29:22,617][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:29:23,116][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:29:23,615][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:29:24,113][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:29:24,613][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:29:25,112][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:29:25,610][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:29:26,108][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:29:26,606][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:29:27,104][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:29:27,603][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:29:28,102][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:29:28,601][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:29:29,100][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:29:29,598][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:29:30,097][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:29:30,596][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:29:31,095][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:29:31,593][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:29:32,091][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:29:32,589][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:29:33,087][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:29:33,585][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:29:34,083][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:29:34,581][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:29:35,080][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:29:35,579][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:29:36,078][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:29:36,577][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:29:37,076][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:29:37,574][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:29:38,073][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:29:38,572][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:29:39,072][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:29:39,570][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:29:40,069][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:29:40,568][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:29:41,065][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:29:41,562][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:29:42,060][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:29:42,558][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 00:29:43,191][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:04 [2026-03-26 00:29:43,932][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:29:43,934][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:29:43,935][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:29:44,806][__main__][INFO] - Iteration 181 took 1m 14s (8.76% Gen, 90.06% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 49m 48s. Estimated total time: 61h 44m 51s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 29s, 500 more iterations: 10h 17m 28s. [2026-03-26 00:29:44,808][__main__][INFO] - Starting iteration 181. [2026-03-26 00:29:45,208][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:29:45,208][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:29:51,455][__main__][INFO] - Number of regex retries in iteration 181: 0 [2026-03-26 00:29:51,456][__main__][INFO] - agents played in iteration 181 are Alice, Bob [2026-03-26 00:29:52,327][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:29:52,372][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:29:52,373][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:29:52,373][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:29:53,250][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:29:53,707][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:29:54,210][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:29:54,709][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:29:55,208][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:29:55,707][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:29:56,205][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:29:56,704][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:29:57,202][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:29:57,701][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:29:58,199][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:29:58,698][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:29:59,196][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:29:59,694][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:30:00,192][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:30:00,690][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:30:01,189][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:30:01,690][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:30:02,190][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:30:02,691][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:30:03,190][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:30:03,688][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:30:04,188][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:30:04,688][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:30:05,189][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:30:05,691][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:30:06,191][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:30:06,688][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:30:07,187][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:30:07,685][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:30:08,183][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:30:08,682][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:30:09,182][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:30:09,680][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:30:10,178][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:30:10,675][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:30:11,174][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:30:11,672][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:30:12,171][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:30:12,670][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:30:13,169][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:30:13,669][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:30:14,167][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:30:14,665][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:30:15,163][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:30:15,661][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:30:16,159][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:30:16,657][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:30:17,157][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:30:17,655][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:30:18,154][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:30:18,651][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:30:19,149][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:30:19,646][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:30:20,144][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:30:20,643][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:30:21,141][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:30:21,639][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:30:22,138][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:30:22,636][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:30:23,134][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:30:23,633][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:30:24,131][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:30:24,630][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:30:25,130][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:30:25,629][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:30:26,127][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:30:26,626][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:30:27,124][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:30:27,623][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:30:28,121][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:30:28,619][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:30:29,118][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:30:29,616][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:30:30,115][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:30:30,613][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:30:31,111][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:30:31,609][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:30:32,107][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:30:32,604][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:30:33,102][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:30:33,600][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:30:34,099][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:30:34,598][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:30:35,097][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:30:35,596][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:30:36,096][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:30:36,596][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:30:37,096][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:30:37,595][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:30:38,093][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:30:38,592][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:30:39,091][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:30:39,589][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:30:40,088][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:30:40,586][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:30:41,084][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:30:41,582][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:30:42,079][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:30:42,577][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:30:43,076][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:30:43,573][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:30:44,071][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:30:44,569][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:30:45,068][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:30:45,566][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:30:46,064][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:30:46,562][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:30:47,059][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:30:47,557][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:30:48,057][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:30:48,557][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:30:49,056][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:30:49,553][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:30:50,053][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:30:50,551][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:30:51,050][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:30:51,548][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:30:52,045][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:30:52,542][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:30:53,039][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:30:53,537][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:30:54,036][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:30:54,534][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:30:55,034][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:30:55,532][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:30:56,030][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:30:56,528][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:30:57,027][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 00:30:57,656][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 00:30:58,392][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:30:58,394][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:30:58,396][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:30:59,229][__main__][INFO] - Iteration 182 took 1m 14s (8.44% Gen, 90.43% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 44m 48s. Estimated total time: 61h 41m 5s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 22s, 500 more iterations: 10h 16m 50s. [2026-03-26 00:30:59,231][__main__][INFO] - Starting iteration 182. [2026-03-26 00:30:59,630][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:30:59,630][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:31:05,989][__main__][INFO] - Number of regex retries in iteration 182: 0 [2026-03-26 00:31:05,990][__main__][INFO] - agents played in iteration 182 are Alice, Bob [2026-03-26 00:31:06,866][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:31:06,910][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:31:06,911][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:31:06,911][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:31:07,509][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:31:07,964][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:31:08,467][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:31:08,966][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:31:09,464][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:31:09,962][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:31:10,460][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:31:10,959][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:31:11,458][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:31:11,955][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:31:12,452][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:31:12,950][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:31:13,447][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:31:13,943][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:31:14,439][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:31:14,936][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:31:15,435][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:31:15,936][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:31:16,435][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:31:16,934][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:31:17,432][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:31:17,930][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:31:18,430][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:31:18,928][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:31:19,426][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:31:19,926][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:31:20,425][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:31:20,925][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:31:21,424][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:31:21,923][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:31:22,421][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:31:22,920][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:31:23,418][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:31:23,917][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:31:24,415][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:31:24,915][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:31:25,414][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:31:25,913][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:31:26,411][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:31:26,909][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:31:27,408][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:31:27,904][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:31:28,401][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:31:28,898][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:31:29,396][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:31:29,893][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:31:30,390][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:31:30,888][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:31:31,386][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:31:31,885][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:31:32,382][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:31:32,881][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:31:33,380][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:31:33,879][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:31:34,378][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:31:34,877][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:31:35,377][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:31:35,877][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:31:36,376][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:31:36,874][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:31:37,374][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:31:37,873][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:31:38,372][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:31:38,871][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:31:39,370][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:31:39,869][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:31:40,367][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:31:40,866][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:31:41,365][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:31:41,862][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:31:42,360][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:31:42,858][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:31:43,357][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:31:43,854][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:31:44,352][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:31:44,850][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:31:45,349][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:31:45,845][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:31:46,341][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:31:46,839][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:31:47,336][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:31:47,835][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:31:48,335][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:31:48,832][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:31:49,330][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:31:49,829][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:31:50,327][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:31:50,825][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:31:51,322][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:31:51,821][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:31:52,319][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:31:52,817][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:31:53,316][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:31:53,815][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:31:54,314][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:31:54,813][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:31:55,313][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:31:55,810][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:31:56,307][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:31:56,805][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:31:57,303][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:31:57,801][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:31:58,299][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:31:58,798][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:31:59,296][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:31:59,793][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:32:00,291][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:32:00,787][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:32:01,284][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:32:01,781][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:32:02,278][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:32:02,777][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:32:03,274][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:32:03,773][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:32:04,272][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:32:04,771][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:32:05,269][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:32:05,767][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:32:06,266][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:32:06,765][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:32:07,263][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:32:07,760][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:32:08,259][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:32:08,757][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:32:09,256][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:32:09,754][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:32:10,253][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:32:10,752][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:32:11,251][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21742 tokens. [2026-03-26 00:32:11,874][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 00:32:12,608][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:32:12,611][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:32:12,613][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:32:13,270][__main__][INFO] - Iteration 183 took 1m 13s (8.64% Gen, 90.47% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 24m 31s. Estimated total time: 61h 22m 3s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 44s, 500 more iterations: 10h 13m 40s. [2026-03-26 00:32:13,272][__main__][INFO] - Starting iteration 183. [2026-03-26 00:32:13,671][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:32:13,672][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:32:18,760][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:32:19,818][__main__][INFO] - Number of regex retries in iteration 183: 1 [2026-03-26 00:32:19,819][__main__][INFO] - agents played in iteration 183 are Alice, Bob [2026-03-26 00:32:20,692][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:32:20,736][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:32:20,737][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:32:20,737][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:32:21,339][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:32:21,793][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:32:22,296][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:32:22,794][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:32:23,292][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:32:23,791][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:32:24,290][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:32:24,789][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:32:25,287][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:32:25,785][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:32:26,282][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:32:26,780][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:32:27,277][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:32:27,775][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:32:28,273][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:32:28,771][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:32:29,269][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:32:29,767][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:32:30,266][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:32:30,764][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:32:31,262][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:32:31,761][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:32:32,260][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:32:32,758][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:32:33,257][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:32:33,755][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:32:34,254][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:32:34,753][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:32:35,251][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:32:35,752][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:32:36,251][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:32:36,750][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:32:37,249][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:32:37,746][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:32:38,245][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:32:38,742][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:32:39,239][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:32:39,737][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:32:40,237][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:32:40,735][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:32:41,233][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:32:41,731][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:32:42,229][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:32:42,727][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:32:43,224][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:32:43,722][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:32:44,220][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:32:44,717][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:32:45,216][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:32:45,718][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:32:46,218][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:32:46,716][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:32:47,214][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:32:47,713][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:32:48,215][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:32:48,715][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:32:49,214][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:32:49,712][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:32:50,210][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:32:50,708][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:32:51,206][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:32:51,705][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:32:52,203][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:32:52,702][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:32:53,200][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:32:53,698][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:32:54,196][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:32:54,694][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:32:55,192][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:32:55,690][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:32:56,190][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:32:56,689][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:32:57,187][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:32:57,685][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:32:58,182][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:32:58,679][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:32:59,176][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:32:59,674][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:33:00,171][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:33:00,669][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:33:01,166][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:33:01,664][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:33:02,162][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:33:02,660][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:33:03,158][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:33:03,656][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:33:04,154][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:33:04,653][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:33:05,153][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:33:05,651][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:33:06,149][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:33:06,647][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:33:07,146][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:33:07,643][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:33:08,140][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:33:08,638][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:33:09,136][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:33:09,633][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:33:10,131][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:33:10,629][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:33:11,126][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:33:11,623][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:33:12,121][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:33:12,618][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:33:13,116][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:33:13,614][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:33:14,112][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:33:14,611][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:33:15,109][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:33:15,606][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:33:16,103][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:33:16,600][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:33:17,097][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:33:17,596][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:33:18,095][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:33:18,595][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:33:19,094][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:33:19,593][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:33:20,091][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:33:20,589][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:33:21,087][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:33:21,585][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:33:22,083][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:33:22,582][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:33:23,081][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:33:23,579][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:33:24,077][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:33:24,575][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:33:25,074][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:33:25,699][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 00:33:26,427][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:33:26,429][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:33:26,431][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:33:27,232][__main__][INFO] - Iteration 184 took 1m 13s (8.36% Gen, 90.55% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 19m 20s. Estimated total time: 61h 18m 5s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 36s, 500 more iterations: 10h 13m 0s. [2026-03-26 00:33:27,235][__main__][INFO] - Starting iteration 184. [2026-03-26 00:33:27,632][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:33:27,633][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:33:33,933][__main__][INFO] - Number of regex retries in iteration 184: 0 [2026-03-26 00:33:33,933][__main__][INFO] - agents played in iteration 184 are Alice, Bob [2026-03-26 00:33:34,807][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:33:34,851][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:33:34,851][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:33:34,852][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:33:35,457][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:33:35,912][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:33:36,416][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:33:36,917][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:33:37,417][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:33:37,917][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:33:38,418][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:33:38,918][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:33:39,417][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:33:39,915][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:33:40,414][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:33:40,913][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:33:41,412][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:33:41,912][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:33:42,411][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:33:42,909][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:33:43,406][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:33:43,905][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:33:44,404][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:33:44,903][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:33:45,402][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:33:45,901][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:33:46,400][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:33:46,898][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:33:47,396][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:33:47,894][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:33:48,393][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:33:48,891][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:33:49,390][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:33:49,888][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:33:50,385][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:33:50,884][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:33:51,381][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:33:51,878][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:33:52,376][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:33:52,875][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:33:53,373][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:33:53,872][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:33:54,371][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:33:54,869][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:33:55,368][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:33:55,865][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:33:56,364][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:33:56,862][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:33:57,360][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:33:57,858][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:33:58,358][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:33:58,858][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:33:59,356][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:33:59,855][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:34:00,353][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:34:00,853][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:34:01,351][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:34:01,852][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:34:02,352][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:34:02,851][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:34:03,350][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:34:03,847][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:34:04,344][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:34:04,840][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:34:05,337][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:34:05,833][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:34:06,331][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:34:06,828][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:34:07,325][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:34:07,823][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:34:08,323][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:34:08,821][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:34:09,321][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:34:09,820][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:34:10,318][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:34:10,817][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:34:11,316][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:34:11,815][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:34:12,312][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:34:12,809][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:34:13,307][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:34:13,805][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:34:14,302][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:34:14,800][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:34:15,299][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:34:15,801][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:34:16,300][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:34:16,798][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:34:17,297][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:34:17,795][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:34:18,295][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:34:18,795][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:34:19,294][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:34:19,792][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:34:20,289][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:34:20,786][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:34:21,283][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:34:21,780][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:34:22,278][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:34:22,774][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:34:23,274][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:34:23,773][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:34:24,271][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:34:24,770][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:34:25,268][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:34:25,766][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:34:26,263][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:34:26,761][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:34:27,260][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:34:27,759][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:34:28,257][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:34:28,755][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:34:29,253][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:34:29,751][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:34:30,249][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:34:30,747][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:34:31,245][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:34:31,743][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:34:32,241][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:34:32,741][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:34:33,239][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:34:33,737][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:34:34,236][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:34:34,734][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:34:35,234][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:34:35,733][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:34:36,231][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:34:36,730][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:34:37,227][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:34:37,723][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:34:38,220][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:34:38,717][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:34:39,215][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:34:39,847][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:34:40,583][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:34:40,585][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:34:40,587][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:34:41,235][__main__][INFO] - Iteration 185 took 1m 13s (8.56% Gen, 90.56% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 20m 11s. Estimated total time: 61h 20m 10s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 40s, 500 more iterations: 10h 13m 21s. [2026-03-26 00:34:41,237][__main__][INFO] - Starting iteration 185. [2026-03-26 00:34:41,635][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:34:41,635][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:34:47,797][__main__][INFO] - Number of regex retries in iteration 185: 0 [2026-03-26 00:34:47,798][__main__][INFO] - agents played in iteration 185 are Alice, Bob [2026-03-26 00:34:48,672][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:34:48,716][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:34:48,717][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:34:48,717][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:34:49,318][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:34:49,773][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:34:50,276][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:34:50,776][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:34:51,275][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:34:51,774][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:34:52,274][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:34:52,773][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:34:53,273][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:34:53,770][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:34:54,266][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:34:54,763][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:34:55,261][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:34:55,760][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:34:56,257][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:34:56,754][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:34:57,252][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:34:57,750][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:34:58,248][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:34:58,746][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:34:59,245][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:34:59,743][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:35:00,241][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:35:00,740][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:35:01,238][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:35:01,737][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:35:02,236][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:35:02,734][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:35:03,233][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:35:03,733][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:35:04,232][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:35:04,731][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:35:05,230][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:35:05,727][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:35:06,227][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:35:06,724][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:35:07,222][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:35:07,719][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:35:08,217][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:35:08,715][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:35:09,214][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:35:09,713][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:35:10,211][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:35:10,711][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:35:11,209][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:35:11,708][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:35:12,207][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:35:12,705][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:35:13,204][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:35:13,703][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:35:14,202][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:35:14,700][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:35:15,200][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:35:15,699][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:35:16,199][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:35:16,698][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:35:17,197][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:35:17,695][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:35:18,193][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:35:18,690][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:35:19,187][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:35:19,684][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:35:20,181][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:35:20,679][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:35:21,176][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:35:21,675][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:35:22,173][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:35:22,671][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:35:23,170][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:35:23,668][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:35:24,166][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:35:24,664][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:35:25,163][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:35:25,660][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:35:26,157][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:35:26,654][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:35:27,151][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:35:27,647][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:35:28,144][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:35:28,640][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:35:29,137][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:35:29,635][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:35:30,133][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:35:30,632][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:35:31,130][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:35:31,628][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:35:32,127][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:35:32,624][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:35:33,123][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:35:33,621][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:35:34,119][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:35:34,618][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:35:35,116][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:35:35,615][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:35:36,112][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:35:36,610][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:35:37,108][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:35:37,605][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:35:38,103][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:35:38,601][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:35:39,101][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:35:39,599][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:35:40,097][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:35:40,595][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:35:41,093][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:35:41,591][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:35:42,092][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:35:42,591][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:35:43,090][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:35:43,590][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:35:44,088][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:35:44,587][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:35:45,085][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:35:45,587][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:35:46,089][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:35:46,588][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:35:47,087][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:35:47,586][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:35:48,086][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:35:48,586][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:35:49,085][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:35:49,582][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:35:50,080][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:35:50,577][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:35:51,076][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:35:51,573][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:35:52,069][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:35:52,566][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:35:53,062][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 00:35:53,690][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 00:35:54,420][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:35:54,423][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:35:54,424][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:35:55,068][__main__][INFO] - Iteration 186 took 1m 13s (8.39% Gen, 90.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 10m 28s. Estimated total time: 61h 11m 41s. Time estimates for 10 more iterations: 12m 14s, 100 more iterations: 2h 2m 23s, 500 more iterations: 10h 11m 56s. [2026-03-26 00:35:55,070][__main__][INFO] - Starting iteration 186. [2026-03-26 00:35:55,468][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:35:55,469][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:36:02,145][__main__][INFO] - Number of regex retries in iteration 186: 0 [2026-03-26 00:36:02,146][__main__][INFO] - agents played in iteration 186 are Alice, Bob [2026-03-26 00:36:03,018][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:36:03,062][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:36:03,063][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:36:03,063][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:36:03,661][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:36:04,116][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:36:04,619][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:36:05,119][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:36:05,618][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:36:06,118][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:36:06,617][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:36:07,116][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:36:07,614][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:36:08,115][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:36:08,616][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:36:09,116][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:36:09,616][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:36:10,116][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:36:10,617][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:36:11,117][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:36:11,619][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:36:12,118][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:36:12,619][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:36:13,119][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:36:13,618][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:36:14,118][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:36:14,616][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:36:15,116][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:36:15,615][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:36:16,115][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:36:16,614][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:36:17,114][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:36:17,613][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:36:18,113][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:36:18,614][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:36:19,114][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:36:19,615][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:36:20,115][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:36:20,615][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:36:21,116][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:36:21,615][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:36:22,115][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:36:22,615][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:36:23,115][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:36:23,615][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:36:24,114][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:36:24,614][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:36:25,112][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:36:25,611][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:36:26,109][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:36:26,608][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:36:27,107][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:36:27,606][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:36:28,104][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:36:28,602][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:36:29,100][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:36:29,599][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:36:30,097][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:36:30,595][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:36:31,093][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:36:31,592][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:36:32,097][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:36:32,614][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:36:33,116][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:36:33,616][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:36:34,117][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:36:34,619][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:36:35,120][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:36:35,620][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:36:36,119][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:36:36,618][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:36:37,118][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:36:37,616][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:36:38,115][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:36:38,614][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:36:39,113][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:36:39,611][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:36:40,112][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:36:40,613][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:36:41,114][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:36:41,613][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:36:42,113][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:36:42,612][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:36:43,113][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:36:43,612][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:36:44,113][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:36:44,613][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:36:45,112][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:36:45,612][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:36:46,111][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:36:46,609][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:36:47,108][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:36:47,606][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:36:48,105][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:36:48,606][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:36:49,105][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:36:49,604][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:36:50,103][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:36:50,601][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:36:51,099][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:36:51,599][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:36:52,099][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:36:52,598][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:36:53,098][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:36:53,598][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:36:54,097][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:36:54,597][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:36:55,096][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:36:55,596][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:36:56,095][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:36:56,594][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:36:57,093][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:36:57,592][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:36:58,091][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:36:58,589][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:36:59,088][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:36:59,588][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:37:00,086][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:37:00,585][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:37:01,082][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:37:01,580][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:37:02,078][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:37:02,577][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:37:03,075][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:37:03,574][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:37:04,076][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:37:04,577][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:37:05,079][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:37:05,580][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:37:06,081][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:37:06,583][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:37:07,083][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:37:07,585][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 00:37:08,205][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 00:37:08,941][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:37:08,943][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:37:08,945][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:37:09,654][__main__][INFO] - Iteration 187 took 1m 14s (9.00% Gen, 90.04% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 46m 51s. Estimated total time: 61h 49m 19s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 38s, 500 more iterations: 10h 18m 13s. [2026-03-26 00:37:09,656][__main__][INFO] - Starting iteration 187. [2026-03-26 00:37:10,053][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:37:10,054][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:37:11,136][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:37:16,276][__main__][INFO] - Number of regex retries in iteration 187: 1 [2026-03-26 00:37:16,277][__main__][INFO] - agents played in iteration 187 are Alice, Bob [2026-03-26 00:37:17,149][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:37:17,194][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:37:17,194][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:37:17,195][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:37:17,808][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:37:18,262][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:37:18,764][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:37:19,262][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:37:19,760][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:37:20,258][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:37:20,756][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:37:21,254][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:37:21,753][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:37:22,254][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:37:22,755][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:37:23,254][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:37:23,754][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:37:24,252][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:37:24,750][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:37:25,249][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:37:25,747][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:37:26,244][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:37:26,742][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:37:27,239][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:37:27,736][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:37:28,234][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:37:28,731][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:37:29,228][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:37:29,726][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:37:30,224][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:37:30,722][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:37:31,220][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:37:31,718][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:37:32,217][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:37:32,715][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:37:33,213][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:37:33,710][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:37:34,208][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:37:34,705][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:37:35,202][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:37:35,700][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:37:36,199][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:37:36,696][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:37:37,194][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:37:37,692][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:37:38,189][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:37:38,686][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:37:39,183][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:37:39,681][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:37:40,179][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:37:40,677][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:37:41,175][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:37:41,674][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:37:42,175][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:37:42,675][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:37:43,176][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:37:43,676][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:37:44,176][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:37:44,676][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:37:45,175][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:37:45,675][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:37:46,173][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:37:46,669][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:37:47,166][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:37:47,663][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:37:48,162][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:37:48,660][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:37:49,156][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:37:49,653][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:37:50,151][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:37:50,648][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:37:51,145][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:37:51,643][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:37:52,140][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:37:52,638][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:37:53,137][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:37:53,635][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:37:54,134][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:37:54,633][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:37:55,131][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:37:55,630][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:37:56,127][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:37:56,625][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:37:57,124][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:37:57,622][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:37:58,120][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:37:58,617][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:37:59,114][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:37:59,611][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:38:00,109][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:38:00,605][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:38:01,103][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:38:01,601][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:38:02,100][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:38:02,599][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:38:03,097][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:38:03,597][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:38:04,096][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:38:04,594][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:38:05,091][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:38:05,591][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:38:06,090][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:38:06,588][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:38:07,086][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:38:07,584][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:38:08,082][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:38:08,580][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:38:09,078][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:38:09,576][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:38:10,076][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:38:10,574][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:38:11,072][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:38:11,570][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:38:12,068][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:38:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:38:13,063][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:38:13,561][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:38:14,063][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:38:14,562][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:38:15,063][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:38:15,562][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:38:16,061][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:38:16,560][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:38:17,059][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:38:17,559][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:38:18,057][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:38:18,556][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:38:19,054][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:38:19,552][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:38:20,049][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:38:20,546][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:38:21,042][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:38:21,538][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 00:38:22,175][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 00:38:22,970][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:38:22,972][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:38:22,973][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:38:23,688][__main__][INFO] - Iteration 188 took 1m 13s (8.45% Gen, 90.58% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 18m 4s. Estimated total time: 61h 21m 46s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 43s, 500 more iterations: 10h 13m 37s. [2026-03-26 00:38:23,690][__main__][INFO] - Starting iteration 188. [2026-03-26 00:38:24,093][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:38:24,094][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:38:31,565][__main__][INFO] - Number of regex retries in iteration 188: 0 [2026-03-26 00:38:31,566][__main__][INFO] - agents played in iteration 188 are Alice, Bob [2026-03-26 00:38:32,442][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:38:32,486][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:38:32,487][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:38:32,487][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:38:33,093][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:38:33,547][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:38:34,050][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:38:34,548][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:38:35,045][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:38:35,542][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:38:36,041][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:38:36,539][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:38:37,037][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:38:37,535][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:38:38,034][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:38:38,532][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:38:39,030][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:38:39,527][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:38:40,025][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:38:40,523][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:38:41,021][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:38:41,522][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:38:42,021][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:38:42,520][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:38:43,019][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:38:43,517][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:38:44,016][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:38:44,515][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:38:45,015][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:38:45,516][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:38:46,018][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:38:46,517][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:38:47,016][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:38:47,515][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:38:48,015][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:38:48,516][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:38:49,015][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:38:49,515][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:38:50,013][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:38:50,511][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:38:51,009][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:38:51,507][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:38:52,005][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:38:52,503][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:38:53,002][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:38:53,504][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:38:54,004][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:38:54,505][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:38:55,004][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:38:55,504][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:38:56,003][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:38:56,502][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:38:57,002][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:38:57,500][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:38:58,000][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:38:58,499][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:38:58,998][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:38:59,497][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:38:59,995][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:39:00,494][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:39:00,992][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:39:01,491][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:39:01,989][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:39:02,487][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:39:02,985][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:39:03,483][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:39:03,982][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:39:04,481][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:39:04,979][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:39:05,478][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:39:05,977][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:39:06,476][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:39:06,974][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:39:07,473][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:39:07,972][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:39:08,471][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:39:08,969][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:39:09,467][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:39:09,965][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:39:10,463][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:39:10,961][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:39:11,459][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:39:11,958][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:39:12,456][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:39:12,955][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:39:13,453][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:39:13,952][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:39:14,451][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:39:14,951][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:39:15,452][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:39:15,953][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:39:16,452][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:39:16,951][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:39:17,449][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:39:17,948][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:39:18,448][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:39:18,946][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:39:19,446][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:39:19,944][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:39:20,443][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:39:20,942][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:39:21,440][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:39:21,938][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:39:22,436][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:39:22,934][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:39:23,435][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:39:23,934][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:39:24,432][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:39:24,931][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:39:25,431][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:39:25,931][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:39:26,430][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:39:26,929][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:39:27,429][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:39:27,928][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:39:28,428][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:39:28,927][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:39:29,425][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:39:29,923][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:39:30,422][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:39:30,920][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:39:31,419][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:39:31,918][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:39:32,417][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:39:32,915][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:39:33,413][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:39:33,911][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:39:34,409][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:39:34,907][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:39:35,405][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:39:35,903][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:39:36,402][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:39:36,901][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 00:39:37,547][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:39:38,277][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:39:38,279][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:39:38,281][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:39:39,070][__main__][INFO] - Iteration 189 took 1m 14s (9.97% Gen, 88.98% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 58h 23m 55s. Estimated total time: 62h 28m 51s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 57s, 500 more iterations: 10h 24m 48s. [2026-03-26 00:39:39,072][__main__][INFO] - Starting iteration 189. [2026-03-26 00:39:39,472][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:39:39,472][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:39:41,031][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:39:45,984][__main__][INFO] - Number of regex retries in iteration 189: 1 [2026-03-26 00:39:45,985][__main__][INFO] - agents played in iteration 189 are Alice, Bob [2026-03-26 00:39:46,860][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:39:46,904][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:39:46,905][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:39:46,905][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:39:47,508][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:39:47,961][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:39:48,472][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:39:48,969][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:39:49,466][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:39:49,963][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:39:50,460][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:39:50,958][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:39:51,457][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:39:51,958][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:39:52,458][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:39:52,958][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:39:53,457][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:39:53,956][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:39:54,455][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:39:54,955][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:39:55,455][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:39:55,952][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:39:56,449][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:39:56,947][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:39:57,443][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:39:57,942][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:39:58,440][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:39:58,937][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:39:59,435][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:39:59,934][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:40:00,430][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:40:00,928][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:40:01,424][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:40:01,922][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:40:02,419][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:40:02,916][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:40:03,414][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:40:03,912][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:40:04,409][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:40:04,906][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:40:05,403][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:40:05,900][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:40:06,397][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:40:06,895][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:40:07,394][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:40:07,894][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:40:08,393][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:40:08,892][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:40:09,391][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:40:09,889][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:40:10,388][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:40:10,886][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:40:11,385][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:40:11,884][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:40:12,384][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:40:12,882][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:40:13,381][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:40:13,879][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:40:14,379][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:40:14,878][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:40:15,378][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:40:15,878][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:40:16,376][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:40:16,874][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:40:17,371][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:40:17,868][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:40:18,367][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:40:18,865][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:40:19,362][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:40:19,861][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:40:20,358][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:40:20,855][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:40:21,351][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:40:21,847][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:40:22,344][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:40:22,841][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:40:23,338][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:40:23,837][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:40:24,337][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:40:24,837][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:40:25,337][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:40:25,836][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:40:26,336][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:40:26,835][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:40:27,334][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:40:27,833][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:40:28,332][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:40:28,829][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:40:29,327][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:40:29,825][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:40:30,322][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:40:30,819][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:40:31,317][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:40:31,815][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:40:32,315][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:40:32,813][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:40:33,312][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:40:33,809][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:40:34,307][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:40:34,805][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:40:35,302][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:40:35,800][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:40:36,298][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:40:36,795][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:40:37,294][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:40:37,793][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:40:38,291][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:40:38,789][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:40:39,286][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:40:39,784][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:40:40,283][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:40:40,782][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:40:41,281][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:40:41,780][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:40:42,280][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:40:42,779][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:40:43,278][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:40:43,777][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:40:44,276][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:40:44,776][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:40:45,275][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:40:45,774][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:40:46,274][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:40:46,771][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:40:47,270][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:40:47,767][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:40:48,267][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:40:48,765][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:40:49,263][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:40:49,760][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:40:50,258][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:40:50,755][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:40:51,252][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 00:40:51,880][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 00:40:52,619][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:40:52,622][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:40:52,623][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:40:53,427][__main__][INFO] - Iteration 190 took 1m 13s (8.81% Gen, 90.10% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 31m 37s. Estimated total time: 61h 37m 48s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 15s, 500 more iterations: 10h 16m 18s. [2026-03-26 00:40:53,429][__main__][INFO] - Starting iteration 190. [2026-03-26 00:40:53,828][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:40:53,829][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:40:54,411][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:40:54,412][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:40:58,117][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:41:00,245][__main__][INFO] - Number of regex retries in iteration 190: 3 [2026-03-26 00:41:00,246][__main__][INFO] - agents played in iteration 190 are Alice, Bob [2026-03-26 00:41:01,121][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:41:01,165][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:41:01,166][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:41:01,167][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:41:01,775][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:41:02,225][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:41:02,722][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:41:03,217][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:41:03,710][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:41:04,204][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:41:04,697][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:41:05,191][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:41:05,685][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:41:06,184][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:41:06,681][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:41:07,180][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:41:07,678][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:41:08,175][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:41:08,674][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:41:09,171][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:41:09,667][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:41:10,167][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:41:10,666][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:41:11,165][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:41:11,664][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:41:12,164][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:41:12,662][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:41:13,161][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:41:13,661][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:41:14,159][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:41:14,657][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:41:15,156][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:41:15,655][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:41:16,154][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:41:16,653][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:41:17,151][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:41:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:41:18,151][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:41:18,654][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:41:19,155][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:41:19,655][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:41:20,154][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:41:20,653][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:41:21,151][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:41:21,650][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:41:22,149][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:41:22,646][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:41:23,145][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:41:23,645][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:41:24,143][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:41:24,642][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:41:25,140][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:41:25,639][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:41:26,138][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:41:26,637][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:41:27,136][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:41:27,634][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:41:28,133][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:41:28,631][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:41:29,130][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:41:29,628][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:41:30,127][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:41:30,626][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:41:31,124][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:41:31,623][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:41:32,121][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:41:32,620][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:41:33,118][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:41:33,617][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:41:34,111][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:41:34,606][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:41:35,101][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:41:35,595][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:41:36,090][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:41:36,584][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:41:37,080][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:41:37,575][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:41:38,074][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:41:38,571][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:41:39,068][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:41:39,566][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:41:40,063][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:41:40,560][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:41:41,059][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:41:41,557][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:41:42,056][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:41:42,555][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:41:43,053][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:41:43,553][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:41:44,052][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:41:44,551][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:41:45,050][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:41:45,552][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:41:46,053][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:41:46,550][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:41:47,048][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:41:47,547][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:41:48,044][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:41:48,542][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:41:49,042][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:41:49,543][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:41:50,042][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:41:50,541][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:41:51,040][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:41:51,539][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:41:52,039][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:41:52,538][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:41:53,037][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:41:53,537][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:41:54,036][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:41:54,534][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:41:55,032][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:41:55,531][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:41:56,029][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:41:56,527][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:41:57,025][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:41:57,524][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:41:58,023][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:41:58,521][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:41:59,020][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:41:59,520][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:42:00,019][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:42:00,518][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:42:01,018][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:42:01,517][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:42:02,016][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:42:02,515][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:42:03,015][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:42:03,513][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:42:04,012][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:42:04,510][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:42:05,009][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:42:05,508][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21739 tokens. [2026-03-26 00:42:06,146][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.30%, ΔTime: 00:01:04 [2026-03-26 00:42:06,888][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:42:06,890][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:42:06,892][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:42:07,551][__main__][INFO] - Iteration 191 took 1m 13s (8.70% Gen, 90.40% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 18m 43s. Estimated total time: 61h 26m 9s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 52s, 500 more iterations: 10h 14m 21s. [2026-03-26 00:42:07,553][__main__][INFO] - Starting iteration 191. [2026-03-26 00:42:07,954][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:42:07,955][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:42:10,141][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:42:14,879][__main__][INFO] - Number of regex retries in iteration 191: 1 [2026-03-26 00:42:14,880][__main__][INFO] - agents played in iteration 191 are Alice, Bob [2026-03-26 00:42:15,760][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:42:15,805][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:42:15,805][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:42:15,806][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:42:16,419][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:42:16,873][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:42:17,377][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:42:17,876][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:42:18,376][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:42:18,875][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:42:19,375][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:42:19,873][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:42:20,370][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:42:20,868][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:42:21,366][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:42:21,864][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:42:22,364][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:42:22,862][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:42:23,361][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:42:23,860][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:42:24,358][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:42:24,859][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:42:25,360][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:42:25,860][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:42:26,359][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:42:26,858][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:42:27,358][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:42:27,857][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:42:28,358][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:42:28,857][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:42:29,357][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:42:29,856][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:42:30,356][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:42:30,855][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:42:31,354][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:42:31,853][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:42:32,352][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:42:32,849][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:42:33,347][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:42:33,844][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:42:34,342][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:42:34,840][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:42:35,338][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:42:35,837][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:42:36,334][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:42:36,833][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:42:37,332][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:42:37,832][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:42:38,331][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:42:38,830][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:42:39,329][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:42:39,827][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:42:40,326][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:42:40,824][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:42:41,322][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:42:41,820][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:42:42,318][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:42:42,816][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:42:43,315][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:42:43,813][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:42:44,310][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:42:44,807][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:42:45,305][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:42:45,804][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:42:46,301][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:42:46,800][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:42:47,298][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:42:47,796][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:42:48,296][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:42:48,793][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:42:49,290][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:42:49,787][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:42:50,284][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:42:50,783][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:42:51,281][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:42:51,778][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:42:52,275][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:42:52,772][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:42:53,270][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:42:53,767][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:42:54,265][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:42:54,763][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:42:55,263][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:42:55,761][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:42:56,259][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:42:56,757][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:42:57,257][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:42:57,757][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:42:58,256][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:42:58,757][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:42:59,256][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:42:59,756][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:43:00,254][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:43:00,752][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:43:01,250][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:43:01,749][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:43:02,246][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:43:02,745][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:43:03,244][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:43:03,742][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:43:04,241][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:43:04,739][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:43:05,236][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:43:05,733][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:43:06,231][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:43:06,729][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:43:07,226][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:43:07,723][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:43:08,221][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:43:08,720][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:43:09,218][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:43:09,716][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:43:10,215][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:43:10,713][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:43:11,211][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:43:11,710][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:43:12,209][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:43:12,707][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:43:13,204][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:43:13,702][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:43:14,200][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:43:14,698][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:43:15,196][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:43:15,698][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:43:16,198][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:43:16,696][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:43:17,195][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:43:17,692][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:43:18,191][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:43:18,688][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:43:19,188][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:43:19,687][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:43:20,185][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 00:43:20,821][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 00:43:21,565][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:43:21,567][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:43:21,569][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:43:22,227][__main__][INFO] - Iteration 192 took 1m 14s (9.32% Gen, 89.79% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 45m 1s. Estimated total time: 61h 53m 41s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 56s. [2026-03-26 00:43:22,229][__main__][INFO] - Starting iteration 192. [2026-03-26 00:43:22,628][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:43:22,628][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:43:29,012][__main__][INFO] - Number of regex retries in iteration 192: 0 [2026-03-26 00:43:29,013][__main__][INFO] - agents played in iteration 192 are Alice, Bob [2026-03-26 00:43:29,885][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:43:29,930][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:43:29,930][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:43:29,931][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:43:30,534][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:43:30,986][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:43:31,486][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:43:31,981][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:43:32,476][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:43:32,970][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:43:33,466][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:43:33,963][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:43:34,458][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:43:34,958][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:43:35,457][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:43:35,956][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:43:36,454][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:43:36,951][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:43:37,448][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:43:37,945][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:43:38,444][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:43:38,944][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:43:39,444][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:43:39,943][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:43:40,442][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:43:40,941][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:43:41,440][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:43:41,939][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:43:42,439][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:43:42,937][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:43:43,436][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:43:43,935][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:43:44,432][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:43:44,931][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:43:45,428][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:43:45,927][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:43:46,424][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:43:46,922][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:43:47,420][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:43:47,919][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:43:48,418][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:43:48,915][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:43:49,413][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:43:49,910][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:43:50,407][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:43:50,905][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:43:51,404][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:43:51,903][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:43:52,402][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:43:52,900][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:43:53,399][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:43:53,897][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:43:54,396][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:43:54,894][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:43:55,393][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:43:55,892][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:43:56,390][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:43:56,887][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:43:57,386][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:43:57,883][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:43:58,382][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:43:58,880][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:43:59,379][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:43:59,877][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:44:00,376][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:44:00,874][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:44:01,372][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:44:01,869][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:44:02,366][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:44:02,862][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:44:03,357][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:44:03,851][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:44:04,344][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:44:04,837][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:44:05,331][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:44:05,825][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:44:06,318][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:44:06,816][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:44:07,313][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:44:07,811][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:44:08,309][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:44:08,806][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:44:09,304][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:44:09,803][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:44:10,302][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:44:10,801][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:44:11,301][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:44:11,799][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:44:12,297][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:44:12,795][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:44:13,293][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:44:13,793][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:44:14,292][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:44:14,790][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:44:15,289][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:44:15,787][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:44:16,284][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:44:16,782][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:44:17,281][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:44:17,779][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:44:18,278][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:44:18,776][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:44:19,276][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:44:19,775][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:44:20,273][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:44:20,770][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:44:21,267][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:44:21,765][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:44:22,264][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:44:22,763][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:44:23,261][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:44:23,760][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:44:24,258][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:44:24,756][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:44:25,255][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:44:25,754][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:44:26,253][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:44:26,751][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:44:27,249][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:44:27,747][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:44:28,245][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:44:28,742][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:44:29,241][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:44:29,739][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:44:30,237][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:44:30,736][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:44:31,235][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:44:31,733][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:44:32,230][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:44:32,727][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:44:33,224][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:44:33,721][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:44:34,218][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:44:34,858][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.26%, ΔTime: 00:01:04 [2026-03-26 00:44:35,590][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:44:35,593][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:44:35,594][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:44:36,419][__main__][INFO] - Iteration 193 took 1m 13s (8.65% Gen, 90.23% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 19m 41s. Estimated total time: 61h 29m 35s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 59s, 500 more iterations: 10h 14m 55s. [2026-03-26 00:44:36,421][__main__][INFO] - Starting iteration 193. [2026-03-26 00:44:36,822][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:44:36,822][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:44:43,799][__main__][INFO] - Number of regex retries in iteration 193: 0 [2026-03-26 00:44:43,799][__main__][INFO] - agents played in iteration 193 are Alice, Bob [2026-03-26 00:44:44,670][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:44:44,996][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:44:44,997][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:44:44,997][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:44:45,611][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:44:46,065][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:44:46,567][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:44:47,066][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:44:47,565][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:44:48,063][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:44:48,562][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:44:49,061][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:44:49,558][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:44:50,060][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:44:50,559][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:44:51,059][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:44:51,557][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:44:52,055][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:44:52,554][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:44:53,053][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:44:53,552][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:44:54,050][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:44:54,548][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:44:55,046][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:44:55,544][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:44:56,042][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:44:56,540][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:44:57,038][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:44:57,535][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:44:58,037][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:44:58,539][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:44:59,038][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:44:59,538][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:45:00,037][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:45:00,536][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:45:01,035][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:45:01,535][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:45:02,034][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:45:02,533][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:45:03,031][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:45:03,528][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:45:04,026][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:45:04,524][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:45:05,022][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:45:05,520][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:45:06,038][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:45:06,542][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:45:07,043][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:45:07,544][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:45:08,045][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:45:08,546][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:45:09,046][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:45:09,546][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:45:10,045][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:45:10,543][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:45:11,042][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:45:11,542][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:45:12,041][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:45:12,540][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:45:13,040][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:45:13,540][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:45:14,037][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:45:14,535][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:45:15,031][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:45:15,528][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:45:16,025][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:45:16,522][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:45:17,019][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:45:17,517][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:45:18,015][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:45:18,515][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:45:19,012][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:45:19,510][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:45:20,008][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:45:20,505][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:45:21,003][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:45:21,501][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:45:22,000][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:45:22,499][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:45:22,997][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:45:23,496][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:45:23,994][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:45:24,493][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:45:24,993][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:45:25,493][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:45:25,991][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:45:26,488][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:45:26,986][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:45:27,484][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:45:27,982][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:45:28,480][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:45:28,978][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:45:29,477][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:45:29,976][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:45:30,477][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:45:30,976][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:45:31,478][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:45:31,977][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:45:32,478][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:45:32,978][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:45:33,478][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:45:33,975][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:45:34,473][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:45:34,970][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:45:35,468][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:45:35,965][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:45:36,463][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:45:36,962][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:45:37,459][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:45:37,960][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:45:38,461][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:45:38,962][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:45:39,463][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:45:39,964][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:45:40,466][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:45:40,968][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:45:41,468][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:45:41,967][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:45:42,466][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:45:42,964][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:45:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:45:43,963][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:45:44,463][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:45:44,962][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:45:45,461][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:45:45,959][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:45:46,455][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:45:46,952][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:45:47,450][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:45:47,947][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:45:48,446][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:45:48,944][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:45:49,441][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 00:45:50,072][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 00:45:51,001][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:45:51,004][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:45:51,005][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:45:52,022][__main__][INFO] - Iteration 194 took 1m 15s (9.28% Gen, 89.37% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 58h 28m 53s. Estimated total time: 62h 40m 3s. Time estimates for 10 more iterations: 12m 32s, 100 more iterations: 2h 5m 20s, 500 more iterations: 10h 26m 40s. [2026-03-26 00:45:52,024][__main__][INFO] - Starting iteration 194. [2026-03-26 00:45:52,422][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:45:52,422][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:45:56,271][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:45:59,023][__main__][INFO] - Number of regex retries in iteration 194: 1 [2026-03-26 00:45:59,024][__main__][INFO] - agents played in iteration 194 are Alice, Bob [2026-03-26 00:45:59,894][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:45:59,938][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:45:59,939][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:45:59,939][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:46:00,543][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:46:00,997][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:46:01,500][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:46:01,998][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:46:02,497][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:46:02,996][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:46:03,495][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:46:03,994][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:46:04,492][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:46:04,995][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:46:05,496][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:46:05,998][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:46:06,501][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:46:07,002][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:46:07,504][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:46:08,006][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:46:08,508][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:46:09,006][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:46:09,504][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:46:10,002][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:46:10,500][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:46:10,999][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:46:11,496][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:46:11,993][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:46:12,491][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:46:12,988][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:46:13,486][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:46:13,984][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:46:14,482][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:46:14,980][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:46:15,482][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:46:15,982][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:46:16,481][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:46:16,980][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:46:17,479][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:46:17,977][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:46:18,476][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:46:18,975][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:46:19,475][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:46:19,973][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:46:20,471][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:46:20,969][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:46:21,466][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:46:21,967][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:46:22,465][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:46:22,964][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:46:23,462][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:46:23,959][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:46:24,456][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:46:24,954][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:46:25,452][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:46:25,951][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:46:26,450][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:46:26,948][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:46:27,446][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:46:27,945][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:46:28,443][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:46:28,942][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:46:29,441][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:46:29,940][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:46:30,439][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:46:30,938][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:46:31,437][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:46:31,936][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:46:32,434][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:46:32,931][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:46:33,429][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:46:33,928][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:46:34,426][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:46:34,925][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:46:35,424][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:46:35,922][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:46:36,421][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:46:36,921][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:46:37,422][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:46:37,924][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:46:38,426][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:46:38,928][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:46:39,429][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:46:39,932][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:46:40,434][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:46:40,932][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:46:41,429][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:46:41,927][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:46:42,425][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:46:42,923][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:46:43,420][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:46:43,917][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:46:44,415][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:46:44,912][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:46:45,410][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:46:45,908][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:46:46,405][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:46:46,902][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:46:47,400][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:46:47,898][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:46:48,397][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:46:48,895][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:46:49,393][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:46:49,892][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:46:50,390][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:46:50,888][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:46:51,386][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:46:51,885][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:46:52,383][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:46:52,880][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:46:53,379][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:46:53,877][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:46:54,375][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:46:54,871][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:46:55,368][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:46:55,865][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:46:56,362][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:46:56,859][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:46:57,357][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:46:57,856][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:46:58,354][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:46:58,852][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:46:59,350][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:46:59,848][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:47:00,346][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:47:00,844][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:47:01,342][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:47:01,840][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:47:02,338][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:47:02,837][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:47:03,336][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:47:03,836][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:47:04,334][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 00:47:04,962][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 00:47:05,701][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:47:05,703][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:47:05,705][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:47:06,517][__main__][INFO] - Iteration 195 took 1m 14s (8.91% Gen, 89.99% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 32m 24s. Estimated total time: 61h 44m 48s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 29s, 500 more iterations: 10h 17m 28s. [2026-03-26 00:47:06,520][__main__][INFO] - Starting iteration 195. [2026-03-26 00:47:06,918][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:47:06,918][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:47:13,528][__main__][INFO] - Number of regex retries in iteration 195: 0 [2026-03-26 00:47:13,529][__main__][INFO] - agents played in iteration 195 are Alice, Bob [2026-03-26 00:47:14,403][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:47:14,447][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:47:14,448][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:47:14,449][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:47:15,047][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:47:15,504][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:47:16,008][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:47:16,506][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:47:17,005][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:47:17,503][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:47:18,003][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:47:18,504][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:47:19,003][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:47:19,510][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:47:20,008][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:47:20,508][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:47:21,006][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:47:21,506][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:47:22,005][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:47:22,505][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:47:23,004][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:47:23,505][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:47:24,005][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:47:24,505][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:47:25,005][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:47:25,506][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:47:26,006][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:47:26,506][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:47:27,006][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:47:27,506][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:47:28,006][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:47:28,505][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:47:29,004][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:47:29,505][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:47:30,004][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:47:30,503][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:47:31,003][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:47:31,503][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:47:32,004][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:47:32,504][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:47:33,003][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:47:33,501][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:47:34,000][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:47:34,499][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:47:34,998][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:47:35,497][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:47:35,996][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:47:36,494][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:47:36,992][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:47:37,491][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:47:37,990][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:47:38,488][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:47:38,988][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:47:39,486][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:47:39,985][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:47:40,483][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:47:40,981][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:47:41,481][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:47:41,979][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:47:42,478][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:47:42,976][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:47:43,477][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:47:43,976][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:47:44,475][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:47:44,975][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:47:45,477][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:47:45,979][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:47:46,479][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:47:46,979][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:47:47,477][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:47:47,976][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:47:48,477][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:47:48,977][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:47:49,477][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:47:49,975][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:47:50,474][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:47:50,972][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:47:51,471][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:47:51,968][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:47:52,466][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:47:52,966][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:47:53,465][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:47:53,963][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:47:54,462][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:47:54,961][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:47:55,462][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:47:55,962][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:47:56,461][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:47:56,961][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:47:57,461][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:47:57,961][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:47:58,460][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:47:58,959][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:47:59,458][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:47:59,957][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:48:00,457][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:48:00,956][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:48:01,455][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:48:01,953][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:48:02,451][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:48:02,949][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:48:03,448][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:48:03,946][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:48:04,444][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:48:04,943][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:48:05,442][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:48:05,941][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:48:06,439][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:48:06,938][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:48:07,436][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:48:07,934][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:48:08,432][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:48:08,931][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:48:09,429][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:48:09,928][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:48:10,426][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:48:10,924][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:48:11,422][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:48:11,921][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:48:12,419][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:48:12,918][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:48:13,417][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:48:13,915][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:48:14,414][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:48:14,911][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:48:15,410][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:48:15,909][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:48:16,407][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:48:16,906][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:48:17,406][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:48:17,906][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:48:18,407][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:48:18,907][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21729 tokens. [2026-03-26 00:48:19,535][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:48:20,275][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:48:20,277][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:48:20,279][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:48:20,972][__main__][INFO] - Iteration 196 took 1m 14s (8.93% Gen, 90.14% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 29m 6s. Estimated total time: 61h 42m 45s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 25s, 500 more iterations: 10h 17m 7s. [2026-03-26 00:48:20,975][__main__][INFO] - Starting iteration 196. [2026-03-26 00:48:21,375][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:48:21,376][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:48:25,599][mllm.models.large_language_model_local][WARNING] - Response Given the updated values where your per-item values are hats=1, books=1, balls=10 and Bob's per-item values are hats=10, books=1, balls=1, it is clear that balls are highly valued by both of us. Therefore, we should propose to take all the balls to maximize our points. Proposal: 0 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:48:28,559][__main__][INFO] - Number of regex retries in iteration 196: 1 [2026-03-26 00:48:28,560][__main__][INFO] - agents played in iteration 196 are Alice, Bob [2026-03-26 00:48:29,440][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:48:29,485][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:48:29,485][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:48:29,486][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:48:30,089][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:48:30,543][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:48:31,044][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:48:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:48:32,040][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:48:32,538][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:48:33,037][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:48:33,534][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:48:34,032][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:48:34,533][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:48:35,032][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:48:35,530][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:48:36,029][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:48:36,528][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:48:37,027][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:48:37,526][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:48:38,026][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:48:38,525][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:48:39,024][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:48:39,523][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:48:40,022][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:48:40,521][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:48:41,020][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:48:41,519][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:48:42,018][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:48:42,518][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:48:43,020][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:48:43,520][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:48:44,020][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:48:44,520][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:48:45,020][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:48:45,520][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:48:46,020][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:48:46,516][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:48:47,012][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:48:47,509][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:48:48,006][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:48:48,506][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:48:49,004][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:48:49,501][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:48:49,997][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:48:50,495][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:48:50,993][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:48:51,490][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:48:51,988][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:48:52,486][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:48:52,984][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:48:53,482][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:48:53,980][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:48:54,479][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:48:54,978][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:48:55,478][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:48:55,977][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:48:56,476][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:48:56,975][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:48:57,474][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:48:57,972][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:48:58,469][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:48:58,967][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:48:59,464][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:48:59,961][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:49:00,459][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:49:00,958][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:49:01,455][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:49:01,952][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:49:02,450][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:49:02,947][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:49:03,445][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:49:03,943][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:49:04,442][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:49:04,940][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:49:05,438][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:49:05,935][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:49:06,435][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:49:06,934][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:49:07,432][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:49:07,930][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:49:08,429][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:49:08,928][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:49:09,427][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:49:09,926][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:49:10,425][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:49:10,924][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:49:11,422][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:49:11,922][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:49:12,421][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:49:12,921][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:49:13,419][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:49:13,918][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:49:14,418][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:49:14,918][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:49:15,420][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:49:15,923][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:49:16,424][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:49:16,924][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:49:17,424][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:49:17,924][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:49:18,423][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:49:18,921][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:49:19,418][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:49:19,915][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:49:20,411][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:49:20,908][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:49:21,405][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:49:21,902][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:49:22,401][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:49:22,898][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:49:23,395][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:49:23,892][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:49:24,391][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:49:24,889][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:49:25,387][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:49:25,885][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:49:26,384][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:49:26,882][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:49:27,381][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:49:27,880][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:49:28,378][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:49:28,878][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:49:29,377][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:49:29,876][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:49:30,373][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:49:30,870][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:49:31,368][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:49:31,865][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:49:32,363][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:49:32,860][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:49:33,359][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:49:33,857][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 00:49:34,489][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 00:49:35,234][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:49:35,269][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:49:35,271][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:49:35,926][__main__][INFO] - Iteration 197 took 1m 14s (9.64% Gen, 89.48% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 57h 52m 43s. Estimated total time: 62h 7m 36s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 15s, 500 more iterations: 10h 21m 16s. [2026-03-26 00:49:35,928][__main__][INFO] - Starting iteration 197. [2026-03-26 00:49:36,328][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:49:36,329][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:49:40,206][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:49:42,876][__main__][INFO] - Number of regex retries in iteration 197: 1 [2026-03-26 00:49:42,877][__main__][INFO] - agents played in iteration 197 are Alice, Bob [2026-03-26 00:49:43,749][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:49:43,793][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:49:43,794][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:49:43,794][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:49:44,399][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:49:44,853][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:49:45,356][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:49:45,855][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:49:46,352][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:49:46,850][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:49:47,348][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:49:47,847][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:49:48,346][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:49:48,849][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:49:49,348][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:49:49,847][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:49:50,345][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:49:50,844][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:49:51,343][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:49:51,842][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:49:52,341][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:49:52,838][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:49:53,337][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:49:53,835][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:49:54,332][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:49:54,828][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:49:55,327][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:49:55,824][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:49:56,321][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:49:56,817][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:49:57,313][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:49:57,809][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:49:58,305][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:49:58,801][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:49:59,298][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:49:59,795][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:50:00,291][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:50:00,790][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:50:01,290][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:50:01,790][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:50:02,290][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:50:02,789][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:50:03,288][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:50:03,787][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:50:04,287][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:50:04,789][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:50:05,289][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:50:05,791][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:50:06,291][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:50:06,790][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:50:07,289][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:50:07,788][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:50:08,288][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:50:08,787][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:50:09,284][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:50:09,783][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:50:10,281][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:50:10,780][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:50:11,279][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:50:11,778][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:50:12,277][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:50:12,775][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:50:13,273][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:50:13,771][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:50:14,269][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:50:14,767][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:50:15,265][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:50:15,764][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:50:16,263][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:50:16,761][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:50:17,260][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:50:17,758][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:50:18,259][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:50:18,759][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:50:19,258][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:50:19,758][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:50:20,257][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:50:20,757][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:50:21,256][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:50:21,755][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:50:22,254][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:50:22,751][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:50:23,249][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:50:23,749][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:50:24,247][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:50:24,745][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:50:25,243][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:50:25,742][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:50:26,240][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:50:26,738][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:50:27,236][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:50:27,733][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:50:28,231][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:50:28,726][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:50:29,222][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:50:29,719][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:50:30,216][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:50:30,712][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:50:31,208][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:50:31,705][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:50:32,202][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:50:32,702][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:50:33,201][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:50:33,701][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:50:34,200][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:50:34,700][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:50:35,199][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:50:35,698][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:50:36,196][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:50:36,697][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:50:37,197][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:50:37,698][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:50:38,197][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:50:38,696][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:50:39,196][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:50:39,695][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:50:40,196][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:50:40,694][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:50:41,192][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:50:41,691][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:50:42,189][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:50:42,686][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:50:43,185][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:50:43,684][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:50:44,182][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:50:44,680][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:50:45,179][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:50:45,681][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:50:46,181][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:50:46,686][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:50:47,186][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:50:47,686][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:50:48,188][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 00:50:48,853][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:50:49,590][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:50:49,592][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:50:49,594][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:50:50,244][__main__][INFO] - Iteration 198 took 1m 13s (8.86% Gen, 90.26% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 19m 40s. Estimated total time: 61h 35m 48s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 11s, 500 more iterations: 10h 15m 58s. [2026-03-26 00:50:50,247][__main__][INFO] - Starting iteration 198. [2026-03-26 00:50:50,655][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:50:50,655][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:50:51,240][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:50:53,922][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:50:55,517][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:50:57,211][__main__][INFO] - Number of regex retries in iteration 198: 3 [2026-03-26 00:50:57,212][__main__][INFO] - agents played in iteration 198 are Alice, Bob [2026-03-26 00:50:58,089][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:50:58,134][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:50:58,134][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:50:58,135][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:50:58,740][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:50:59,194][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:50:59,696][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:51:00,203][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:51:00,702][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:51:01,203][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:51:01,702][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:51:02,202][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:51:02,700][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:51:03,200][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:51:03,699][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:51:04,197][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:51:04,698][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:51:05,198][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:51:05,698][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:51:06,206][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:51:06,705][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:51:07,206][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:51:07,706][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:51:08,206][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:51:08,706][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:51:09,205][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:51:09,705][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:51:10,204][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:51:10,706][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:51:11,207][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:51:11,708][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:51:12,208][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:51:12,708][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:51:13,208][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:51:13,708][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:51:14,207][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:51:14,707][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:51:15,206][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:51:15,705][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:51:16,205][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:51:16,704][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:51:17,203][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:51:17,702][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:51:18,204][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:51:18,704][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:51:19,202][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:51:19,700][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:51:20,198][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:51:20,695][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:51:21,193][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:51:21,690][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:51:22,187][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:51:22,685][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:51:23,184][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:51:23,683][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:51:24,182][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:51:24,681][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:51:25,180][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:51:25,679][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:51:26,178][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:51:26,678][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:51:27,178][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:51:27,679][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:51:28,179][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:51:28,677][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:51:29,176][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:51:29,674][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:51:30,173][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:51:30,671][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:51:31,168][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:51:31,666][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:51:32,164][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:51:32,660][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:51:33,158][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:51:33,656][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:51:34,153][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:51:34,650][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:51:35,147][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:51:35,646][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:51:36,145][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:51:36,644][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:51:37,141][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:51:37,640][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:51:38,138][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:51:38,636][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:51:39,134][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:51:39,634][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:51:40,133][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:51:40,632][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:51:41,130][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:51:41,630][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:51:42,129][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:51:42,628][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:51:43,128][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:51:43,628][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:51:44,128][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:51:44,627][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:51:45,127][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:51:45,627][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:51:46,126][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:51:46,625][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:51:47,124][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:51:47,623][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:51:48,121][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:51:48,621][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:51:49,120][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:51:49,619][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:51:50,116][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:51:50,614][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:51:51,112][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:51:51,610][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:51:52,108][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:51:52,606][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:51:53,104][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:51:53,601][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:51:54,099][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:51:54,595][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:51:55,094][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:51:55,593][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:51:56,090][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:51:56,588][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:51:57,085][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:51:57,583][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:51:58,082][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:51:58,581][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:51:59,080][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:51:59,579][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:52:00,079][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:52:00,579][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:52:01,079][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:52:01,577][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:52:02,075][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:52:02,573][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 00:52:03,224][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:52:03,973][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:52:03,975][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:52:03,976][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:52:04,681][__main__][INFO] - Iteration 199 took 1m 14s (8.86% Gen, 90.19% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 23m 57s. Estimated total time: 61h 41m 20s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 22s, 500 more iterations: 10h 16m 53s. [2026-03-26 00:52:04,683][__main__][INFO] - Starting iteration 199. [2026-03-26 00:52:05,083][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:52:05,084][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:52:06,241][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:52:11,581][__main__][INFO] - Number of regex retries in iteration 199: 1 [2026-03-26 00:52:11,581][__main__][INFO] - agents played in iteration 199 are Alice, Bob [2026-03-26 00:52:12,464][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:52:12,509][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:52:12,509][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:52:12,510][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:52:13,124][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:52:13,579][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:52:14,084][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:52:14,583][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:52:15,083][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:52:15,585][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:52:16,087][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:52:16,586][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:52:17,085][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:52:17,585][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:52:18,084][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:52:18,586][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:52:19,085][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:52:19,584][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:52:20,083][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:52:20,582][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:52:21,081][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:52:21,582][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:52:22,083][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:52:22,584][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:52:23,084][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:52:23,586][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:52:24,086][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:52:24,586][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:52:25,087][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:52:25,586][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:52:26,084][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:52:26,583][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:52:27,082][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:52:27,581][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:52:28,080][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:52:28,579][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:52:29,077][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:52:29,576][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:52:30,074][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:52:30,572][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:52:31,070][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:52:31,570][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:52:32,069][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:52:32,568][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:52:33,068][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:52:33,566][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:52:34,065][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:52:34,564][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:52:35,062][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:52:35,561][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:52:36,060][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:52:36,560][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:52:37,058][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:52:37,556][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:52:38,053][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:52:38,550][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:52:39,047][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:52:39,546][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:52:40,045][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:52:40,543][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:52:41,041][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:52:41,540][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:52:42,040][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:52:42,540][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:52:43,040][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:52:43,539][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:52:44,038][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:52:44,536][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:52:45,035][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:52:45,533][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:52:46,033][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:52:46,532][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:52:47,030][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:52:47,529][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:52:48,029][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:52:48,530][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:52:49,029][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:52:49,527][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:52:50,026][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:52:50,524][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:52:51,023][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:52:51,521][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:52:52,020][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:52:52,519][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:52:53,017][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:52:53,516][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:52:54,016][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:52:54,516][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:52:55,016][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:52:55,517][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:52:56,017][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:52:56,518][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:52:57,017][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:52:57,516][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:52:58,013][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:52:58,510][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:52:59,008][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:52:59,506][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:53:00,004][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:53:00,502][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:53:01,001][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:53:01,500][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:53:01,999][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:53:02,497][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:53:02,996][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:53:03,495][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:53:03,994][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:53:04,493][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:53:04,992][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:53:05,489][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:53:05,988][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:53:06,486][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:53:06,986][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:53:07,485][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:53:07,983][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:53:08,481][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:53:08,980][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:53:09,477][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:53:09,975][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:53:10,473][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:53:10,970][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:53:11,468][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:53:11,967][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:53:12,465][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:53:12,963][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:53:13,461][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:53:13,959][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:53:14,458][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:53:14,956][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:53:15,454][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:53:15,952][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:53:16,450][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:53:16,949][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21723 tokens. [2026-03-26 00:53:17,592][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 00:53:18,332][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:53:18,335][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:53:18,336][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:53:18,995][__main__][INFO] - Iteration 200 took 1m 13s (8.79% Gen, 90.32% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 16m 58s. Estimated total time: 61h 35m 35s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 11s, 500 more iterations: 10h 15m 55s. [2026-03-26 00:53:18,997][__main__][INFO] - Starting iteration 200. [2026-03-26 00:53:19,397][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 3 and human policies 1. [2026-03-26 00:53:19,397][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:53:25,839][__main__][INFO] - Number of regex retries in iteration 200: 0 [2026-03-26 00:53:25,840][__main__][INFO] - agents played in iteration 200 are Alice, Bob [2026-03-26 00:53:26,712][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:53:26,757][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:53:26,757][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:53:26,758][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:53:27,365][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:53:27,819][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:53:28,321][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:53:28,820][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:53:29,321][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:53:29,820][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:53:30,320][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:53:30,820][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:53:31,318][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:53:31,816][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:53:32,312][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:53:32,810][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:53:33,307][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:53:33,805][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:53:34,303][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:53:34,801][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:53:35,299][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:53:35,799][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:53:36,299][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:53:36,799][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:53:37,300][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:53:37,800][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:53:38,300][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:53:38,800][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:53:39,301][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:53:39,800][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:53:40,298][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:53:40,798][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:53:41,297][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:53:41,795][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:53:42,294][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:53:42,793][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:53:43,291][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:53:43,790][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:53:44,288][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:53:44,787][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:53:45,287][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:53:45,790][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:53:46,289][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:53:46,788][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:53:47,287][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:53:47,786][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:53:48,284][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:53:48,785][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:53:49,284][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:53:49,782][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:53:50,280][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:53:50,778][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:53:51,276][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:53:51,775][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:53:52,274][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:53:52,772][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:53:53,270][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:53:53,769][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:53:54,268][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:53:54,766][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:53:55,266][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:53:55,765][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:53:56,264][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:53:56,762][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:53:57,261][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:53:57,760][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:53:58,259][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:53:58,758][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:53:59,256][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:53:59,753][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:54:00,251][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:54:00,749][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:54:01,247][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:54:01,745][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:54:02,243][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:54:02,741][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:54:03,239][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:54:03,736][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:54:04,232][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:54:04,729][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:54:05,227][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:54:05,724][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:54:06,222][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:54:06,720][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:54:07,218][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:54:07,717][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:54:08,217][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:54:08,716][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:54:09,215][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:54:09,715][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:54:10,214][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:54:10,714][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:54:11,214][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:54:11,712][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:54:12,210][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:54:12,708][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:54:13,207][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:54:13,706][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:54:14,205][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:54:14,704][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:54:15,203][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:54:15,701][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:54:16,199][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:54:16,697][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:54:17,196][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:54:17,695][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:54:18,196][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:54:18,696][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:54:19,194][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:54:19,693][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:54:20,191][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:54:20,688][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:54:21,187][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:54:21,686][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:54:22,184][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:54:22,683][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:54:23,181][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:54:23,680][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:54:24,179][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:54:24,678][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:54:25,179][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:54:25,679][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:54:26,178][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:54:26,677][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:54:27,175][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:54:27,673][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:54:28,171][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:54:28,669][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:54:29,168][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:54:29,667][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:54:30,166][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:54:30,666][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:54:31,165][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:54:31,799][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 00:54:32,529][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:54:32,531][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:54:32,533][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:54:33,767][__main__][INFO] - Iteration 201 took 1m 14s (8.66% Gen, 89.68% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 38m 39s. Estimated total time: 61h 58m 31s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 57s, 500 more iterations: 10h 19m 45s. [2026-03-26 00:54:33,769][__main__][INFO] - Starting iteration 201. [2026-03-26 00:54:34,169][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 00:54:34,170][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:54:34,774][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:54:40,694][__main__][INFO] - Number of regex retries in iteration 201: 1 [2026-03-26 00:54:40,695][__main__][INFO] - agents played in iteration 201 are Alice, Bob [2026-03-26 00:54:41,567][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:54:41,611][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:54:41,612][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:54:41,613][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:54:42,214][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:54:42,671][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:54:43,175][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:54:43,675][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:54:44,174][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:54:44,673][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:54:45,172][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:54:45,671][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:54:46,170][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:54:46,669][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:54:47,170][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:54:47,670][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:54:48,169][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:54:48,669][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:54:49,168][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:54:49,667][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:54:50,166][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:54:50,664][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:54:51,162][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:54:51,660][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:54:52,158][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:54:52,655][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:54:53,152][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:54:53,649][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:54:54,146][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:54:54,643][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:54:55,142][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:54:55,640][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:54:56,138][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:54:56,635][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:54:57,132][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:54:57,629][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:54:58,127][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:54:58,624][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:54:59,122][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:54:59,621][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:55:00,120][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:55:00,618][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:55:01,116][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:55:01,613][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:55:02,111][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:55:02,609][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:55:03,108][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:55:03,606][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:55:04,105][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:55:04,604][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:55:05,102][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:55:05,601][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:55:06,099][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:55:06,598][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:55:07,096][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:55:07,595][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:55:08,093][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:55:08,591][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:55:09,089][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:55:09,587][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:55:10,086][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:55:10,585][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:55:11,083][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:55:11,581][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:55:12,078][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:55:12,576][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:55:13,074][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:55:13,572][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:55:14,069][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:55:14,567][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:55:15,066][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:55:15,568][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:55:16,070][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:55:16,570][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:55:17,070][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:55:17,569][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:55:18,068][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:55:18,568][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:55:19,066][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:55:19,564][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:55:20,063][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:55:20,562][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:55:21,060][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:55:21,559][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:55:22,058][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:55:22,555][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:55:23,053][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:55:23,550][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:55:24,048][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:55:24,547][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:55:25,046][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:55:25,544][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:55:26,041][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:55:26,538][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:55:27,036][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:55:27,534][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:55:28,031][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:55:28,529][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:55:29,028][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:55:29,526][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:55:30,024][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:55:30,522][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:55:31,019][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:55:31,518][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:55:32,016][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:55:32,514][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:55:33,014][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:55:33,512][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:55:34,010][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:55:34,509][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:55:35,007][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:55:35,506][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:55:36,004][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:55:36,503][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:55:37,002][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:55:37,501][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:55:38,000][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:55:38,499][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:55:38,998][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:55:39,496][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:55:39,994][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:55:40,492][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:55:40,991][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:55:41,489][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:55:41,988][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:55:42,487][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:55:42,985][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:55:43,483][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:55:43,981][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:55:44,479][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:55:44,976][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:55:45,474][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:55:45,971][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:55:46,595][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 00:55:47,335][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:55:47,337][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:55:47,339][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:55:48,001][__main__][INFO] - Iteration 202 took 1m 13s (8.84% Gen, 90.26% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 10m 30s. Estimated total time: 61h 31m 36s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 3s, 500 more iterations: 10h 15m 16s. [2026-03-26 00:55:48,003][__main__][INFO] - Starting iteration 202. [2026-03-26 00:55:48,406][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 00:55:48,407][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:55:49,009][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:55:55,178][__main__][INFO] - Number of regex retries in iteration 202: 1 [2026-03-26 00:55:55,178][__main__][INFO] - agents played in iteration 202 are Alice, Bob [2026-03-26 00:55:56,065][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:55:56,110][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:55:56,110][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:55:56,111][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:55:56,724][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:55:57,179][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:55:57,681][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:55:58,180][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:55:58,678][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:55:59,176][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:55:59,674][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:56:00,172][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:56:00,670][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:56:01,171][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:56:01,669][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:56:02,168][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:56:02,667][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:56:03,166][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:56:03,665][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:56:04,164][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:56:04,663][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:56:05,161][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:56:05,659][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:56:06,155][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:56:06,652][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:56:07,149][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:56:07,646][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:56:08,143][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:56:08,642][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:56:09,140][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:56:09,638][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:56:10,135][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:56:10,633][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:56:11,129][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:56:11,627][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:56:12,125][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:56:12,623][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:56:13,124][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:56:13,624][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:56:14,124][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:56:14,623][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:56:15,122][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:56:15,621][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:56:16,120][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:56:16,621][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:56:17,120][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:56:17,646][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:56:18,146][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:56:18,646][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:56:19,145][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:56:19,644][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:56:20,142][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:56:20,641][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:56:21,140][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:56:21,639][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:56:22,137][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:56:22,637][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:56:23,136][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:56:23,635][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:56:24,134][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:56:24,633][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:56:25,132][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:56:25,631][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:56:26,682][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:56:27,178][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:56:27,677][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:56:28,175][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:56:28,674][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:56:29,173][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:56:29,671][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:56:30,169][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:56:30,667][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:56:31,166][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:56:31,664][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:56:32,164][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:56:32,664][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:56:33,163][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:56:33,663][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:56:34,162][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:56:34,660][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:56:35,159][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:56:35,657][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:56:36,156][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:56:36,654][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:56:37,153][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:56:37,650][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:56:38,147][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:56:38,645][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:56:39,141][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:56:39,638][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:56:40,135][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:56:40,632][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:56:41,130][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:56:41,628][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:56:42,126][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:56:42,624][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:56:43,123][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:56:43,621][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:56:44,118][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:56:44,617][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:56:45,115][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:56:45,618][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:56:46,120][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:56:46,619][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:56:47,118][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:56:47,616][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:56:48,115][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:56:48,616][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:56:49,116][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:56:49,615][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:56:50,113][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:56:50,611][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:56:51,110][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:56:51,608][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:56:52,107][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:56:52,606][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:56:53,104][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:56:53,603][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:56:54,102][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:56:54,601][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:56:55,101][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:56:55,601][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:56:56,102][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:56:56,602][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:56:57,102][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:56:57,601][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:56:58,100][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:56:58,597][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:56:59,095][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:56:59,593][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:57:00,092][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:57:00,590][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:57:01,093][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 00:57:01,749][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:05 [2026-03-26 00:57:02,515][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:57:02,518][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:57:02,519][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:57:03,257][__main__][INFO] - Iteration 203 took 1m 14s (9.05% Gen, 89.97% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 58h 0m 14s. Estimated total time: 62h 22m 35s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 45s, 500 more iterations: 10h 23m 45s. [2026-03-26 00:57:03,262][__main__][INFO] - Starting iteration 203. [2026-03-26 00:57:03,662][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 00:57:03,662][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:57:04,349][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:57:04,367][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:57:06,392][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:57:10,292][__main__][INFO] - Number of regex retries in iteration 203: 3 [2026-03-26 00:57:10,293][__main__][INFO] - agents played in iteration 203 are Alice, Bob [2026-03-26 00:57:11,259][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:57:11,303][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:57:11,304][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:57:11,305][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:57:11,971][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:57:12,425][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:57:12,927][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:57:13,426][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:57:13,924][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:57:14,423][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:57:14,920][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:57:15,421][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:57:15,920][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:57:16,418][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:57:16,915][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:57:17,412][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:57:17,910][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:57:18,410][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:57:18,907][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:57:19,404][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:57:19,903][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:57:20,403][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:57:20,902][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:57:21,401][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:57:21,899][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:57:22,398][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:57:22,895][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:57:23,393][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:57:23,891][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:57:24,389][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:57:24,888][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:57:25,388][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:57:25,886][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:57:26,386][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:57:26,885][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:57:27,384][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:57:27,882][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:57:28,379][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:57:28,875][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:57:29,371][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:57:29,867][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:57:30,363][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:57:30,860][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:57:31,356][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:57:31,852][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:57:32,353][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:57:32,853][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:57:33,352][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:57:33,854][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:57:34,354][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:57:34,855][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:57:35,355][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:57:35,854][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:57:36,351][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:57:36,849][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:57:37,347][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:57:37,844][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:57:38,340][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:57:38,863][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:57:39,361][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:57:39,860][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:57:40,357][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:57:40,855][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:57:41,352][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:57:41,850][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:57:42,347][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:57:42,846][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:57:43,344][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:57:44,041][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:57:44,540][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:57:45,039][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:57:45,538][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:57:46,036][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:57:46,534][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:57:47,033][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:57:47,532][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:57:48,031][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:57:48,530][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:57:49,029][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:57:49,527][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:57:50,024][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:57:50,521][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:57:51,019][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:57:51,516][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:57:52,013][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:57:52,511][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:57:53,010][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:57:53,509][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:57:54,008][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:57:54,507][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:57:55,007][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:57:55,506][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:57:56,004][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:57:56,502][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:57:57,003][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:57:57,503][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:57:58,002][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:57:58,501][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:57:59,000][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:57:59,499][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:57:59,997][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:58:00,492][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:58:00,990][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:58:01,486][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:58:01,984][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:58:02,481][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:58:02,978][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:58:03,474][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:58:03,970][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:58:04,470][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:58:04,970][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:58:05,470][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:58:05,970][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:58:06,470][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:58:06,970][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:58:07,470][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:58:07,969][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:58:08,467][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:58:08,965][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:58:09,464][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:58:09,961][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:58:10,459][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:58:10,956][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:58:11,454][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:58:11,951][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:58:12,449][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:58:12,948][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:58:13,447][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:58:13,946][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:58:14,444][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:58:14,942][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:58:15,443][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:58:15,944][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 00:58:16,602][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 00:58:17,242][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:58:17,244][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:58:17,245][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:58:17,981][__main__][INFO] - Iteration 204 took 1m 14s (8.92% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 32m 23s. Estimated total time: 61h 55m 59s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 51s, 500 more iterations: 10h 19m 19s. [2026-03-26 00:58:17,985][__main__][INFO] - Starting iteration 204. [2026-03-26 00:58:18,392][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 00:58:18,392][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:58:20,014][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 00:58:24,794][__main__][INFO] - Number of regex retries in iteration 204: 1 [2026-03-26 00:58:24,795][__main__][INFO] - agents played in iteration 204 are Alice, Bob [2026-03-26 00:58:25,681][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:58:25,726][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:58:25,726][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:58:25,727][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:58:26,341][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:58:26,794][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:58:27,296][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:58:27,794][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:58:28,293][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:58:28,791][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:58:29,290][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:58:29,788][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:58:30,287][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:58:30,786][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:58:31,286][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:58:31,784][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:58:32,284][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:58:32,782][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:58:33,281][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:58:33,780][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:58:34,279][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:58:34,778][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:58:35,278][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:58:35,777][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:58:36,277][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:58:36,776][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:58:37,274][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:58:37,773][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:58:38,272][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:58:38,770][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:58:39,269][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:58:39,768][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:58:40,268][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:58:40,768][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:58:41,267][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:58:41,765][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:58:42,264][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:58:42,763][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:58:43,261][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:58:43,760][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:58:44,257][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:58:44,755][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:58:45,253][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:58:45,750][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 00:58:46,247][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 00:58:46,746][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 00:58:47,246][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 00:58:47,744][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 00:58:48,243][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 00:58:48,743][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 00:58:49,242][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 00:58:49,740][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 00:58:50,237][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 00:58:50,738][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 00:58:51,236][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 00:58:51,734][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 00:58:52,233][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 00:58:52,732][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 00:58:53,231][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 00:58:53,730][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 00:58:54,228][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 00:58:54,725][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 00:58:55,223][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 00:58:55,721][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 00:58:56,219][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 00:58:56,718][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 00:58:57,216][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 00:58:57,714][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 00:58:58,211][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 00:58:58,709][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 00:58:59,206][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 00:58:59,704][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 00:59:00,203][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 00:59:00,702][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 00:59:01,202][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 00:59:01,700][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 00:59:02,199][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 00:59:02,698][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 00:59:03,195][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 00:59:03,693][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 00:59:04,192][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 00:59:04,692][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 00:59:05,190][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 00:59:05,690][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 00:59:06,188][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 00:59:06,687][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 00:59:07,186][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 00:59:07,685][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 00:59:08,184][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 00:59:08,683][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 00:59:09,182][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 00:59:09,681][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 00:59:10,181][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 00:59:10,679][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 00:59:11,177][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 00:59:11,674][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 00:59:12,172][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 00:59:12,671][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 00:59:13,170][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 00:59:13,669][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 00:59:14,167][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 00:59:14,665][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 00:59:15,163][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 00:59:15,660][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 00:59:16,158][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 00:59:16,656][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 00:59:17,153][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 00:59:17,650][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 00:59:18,150][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 00:59:18,648][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 00:59:19,146][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 00:59:19,645][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 00:59:20,143][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 00:59:20,642][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 00:59:21,142][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 00:59:21,642][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 00:59:22,141][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 00:59:22,641][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 00:59:23,140][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 00:59:23,639][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 00:59:24,137][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 00:59:24,637][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 00:59:25,137][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 00:59:25,636][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 00:59:26,135][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 00:59:26,633][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 00:59:27,131][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 00:59:27,629][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 00:59:28,127][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 00:59:28,626][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 00:59:29,125][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 00:59:29,623][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 00:59:30,122][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 00:59:30,802][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 00:59:31,540][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 00:59:31,542][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 00:59:31,543][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 00:59:32,151][__main__][INFO] - Iteration 205 took 1m 13s (8.68% Gen, 90.49% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 3m 9s. Estimated total time: 61h 27m 59s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 55s, 500 more iterations: 10h 14m 39s. [2026-03-26 00:59:32,153][__main__][INFO] - Starting iteration 205. [2026-03-26 00:59:32,553][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 00:59:32,554][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 00:59:39,085][__main__][INFO] - Number of regex retries in iteration 205: 0 [2026-03-26 00:59:39,086][__main__][INFO] - agents played in iteration 205 are Alice, Bob [2026-03-26 00:59:39,959][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:59:40,003][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 00:59:40,004][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 00:59:40,004][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 00:59:40,611][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 00:59:41,069][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 00:59:41,572][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 00:59:42,070][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 00:59:42,568][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 00:59:43,067][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 00:59:43,565][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 00:59:44,063][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 00:59:44,561][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 00:59:45,055][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 00:59:45,552][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 00:59:46,048][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 00:59:46,542][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 00:59:47,036][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 00:59:47,530][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 00:59:48,024][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 00:59:48,522][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 00:59:49,023][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 00:59:49,522][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 00:59:50,021][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 00:59:50,520][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 00:59:51,018][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 00:59:51,516][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 00:59:52,014][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 00:59:52,514][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 00:59:53,012][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 00:59:53,510][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 00:59:54,008][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 00:59:54,507][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 00:59:55,006][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 00:59:55,504][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 00:59:56,003][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 00:59:56,502][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 00:59:57,001][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 00:59:57,498][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 00:59:57,996][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 00:59:58,495][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 00:59:58,992][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 00:59:59,489][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 00:59:59,987][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:00:00,486][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:00:00,984][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:00:01,482][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:00:01,980][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:00:02,478][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:00:02,976][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:00:03,473][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:00:03,970][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:00:04,468][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:00:04,967][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:00:05,464][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:00:05,962][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:00:06,459][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:00:06,956][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:00:07,453][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:00:07,950][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:00:08,447][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:00:08,946][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:00:09,444][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:00:09,943][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:00:10,442][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:00:10,940][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:00:11,438][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:00:11,936][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:00:12,434][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:00:12,931][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:00:13,430][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:00:13,928][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:00:14,425][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:00:14,923][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:00:15,421][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:00:15,919][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:00:16,417][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:00:16,911][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:00:17,406][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:00:17,902][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:00:18,399][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:00:18,893][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:00:19,388][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:00:19,882][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:00:20,377][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:00:20,875][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:00:21,374][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:00:21,872][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:00:22,370][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:00:22,869][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:00:23,367][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:00:23,866][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:00:24,365][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:00:24,863][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:00:25,361][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:00:25,859][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:00:26,357][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:00:26,855][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:00:27,352][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:00:27,850][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:00:28,348][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:00:28,846][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:00:29,344][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:00:29,841][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:00:30,339][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:00:30,837][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:00:31,333][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:00:31,831][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:00:32,329][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:00:32,827][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:00:33,325][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:00:33,822][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:00:34,321][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:00:34,819][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:00:35,316][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:00:35,815][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:00:36,314][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:00:36,811][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:00:37,308][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:00:37,805][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:00:38,303][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:00:38,800][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:00:39,298][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:00:39,795][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:00:40,291][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:00:40,790][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:00:41,290][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:00:41,789][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:00:42,288][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:00:42,786][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:00:43,284][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:00:43,783][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:00:44,282][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 01:00:44,912][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 01:00:45,679][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:00:45,682][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:00:45,683][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:00:46,331][__main__][INFO] - Iteration 206 took 1m 13s (8.85% Gen, 90.27% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 2m 51s. Estimated total time: 61h 28m 55s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 57s, 500 more iterations: 10h 14m 49s. [2026-03-26 01:00:46,333][__main__][INFO] - Starting iteration 206. [2026-03-26 01:00:46,735][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:00:46,736][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:00:50,212][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:00:53,486][__main__][INFO] - Number of regex retries in iteration 206: 1 [2026-03-26 01:00:53,487][__main__][INFO] - agents played in iteration 206 are Alice, Bob [2026-03-26 01:00:54,358][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:00:54,403][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:00:54,403][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:00:54,404][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:00:55,005][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:00:55,468][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:00:55,970][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:00:56,469][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:00:56,968][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:00:57,468][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:00:57,966][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:00:58,465][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:00:58,964][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:00:59,461][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:00:59,959][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:01:00,457][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:01:00,954][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:01:01,452][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:01:01,951][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:01:02,449][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:01:02,947][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:01:03,446][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:01:03,945][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:01:04,442][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:01:04,940][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:01:05,438][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:01:05,936][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:01:06,434][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:01:06,932][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:01:07,431][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:01:07,929][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:01:08,426][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:01:08,924][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:01:09,422][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:01:09,919][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:01:10,417][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:01:10,915][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:01:11,415][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:01:11,914][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:01:12,413][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:01:12,911][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:01:13,410][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:01:13,909][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:01:14,408][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:01:14,908][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:01:15,407][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:01:15,908][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:01:16,406][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:01:16,904][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:01:17,402][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:01:17,901][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:01:18,402][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:01:18,901][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:01:19,399][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:01:19,896][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:01:20,393][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:01:20,890][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:01:21,388][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:01:21,886][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:01:22,384][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:01:22,882][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:01:23,381][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:01:23,879][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:01:24,377][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:01:24,875][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:01:25,373][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:01:25,871][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:01:26,370][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:01:26,869][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:01:27,370][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:01:27,870][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:01:28,369][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:01:28,869][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:01:29,369][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:01:29,868][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:01:30,368][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:01:30,867][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:01:31,365][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:01:31,863][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:01:32,360][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:01:32,858][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:01:33,355][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:01:33,853][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:01:34,351][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:01:34,849][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:01:35,347][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:01:35,846][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:01:36,344][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:01:36,843][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:01:37,342][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:01:37,842][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:01:38,344][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:01:38,843][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:01:39,341][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:01:39,838][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:01:40,335][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:01:40,834][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:01:41,331][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:01:41,829][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:01:42,327][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:01:42,826][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:01:43,324][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:01:43,823][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:01:44,323][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:01:44,821][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:01:45,320][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:01:45,817][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:01:46,315][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:01:46,813][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:01:47,312][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:01:47,811][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:01:48,312][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:01:48,811][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:01:49,309][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:01:49,808][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:01:50,307][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:01:50,806][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:01:51,304][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:01:51,802][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:01:52,300][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:01:52,797][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:01:53,295][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:01:53,791][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:01:54,289][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:01:54,787][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:01:55,286][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:01:55,785][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:01:56,284][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:01:56,783][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:01:57,281][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:01:57,779][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:01:58,277][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:01:58,775][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 01:01:59,404][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:02:00,135][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:02:00,137][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:02:00,139][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:02:00,789][__main__][INFO] - Iteration 207 took 1m 14s (9.12% Gen, 90.00% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 15m 24s. Estimated total time: 61h 42m 43s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 25s, 500 more iterations: 10h 17m 7s. [2026-03-26 01:02:00,791][__main__][INFO] - Starting iteration 207. [2026-03-26 01:02:01,192][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:02:01,192][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:02:07,878][__main__][INFO] - Number of regex retries in iteration 207: 0 [2026-03-26 01:02:07,879][__main__][INFO] - agents played in iteration 207 are Alice, Bob [2026-03-26 01:02:08,757][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:02:08,801][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:02:08,802][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:02:08,802][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:02:09,411][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:02:09,866][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:02:10,369][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:02:10,868][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:02:11,367][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:02:11,866][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:02:12,366][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:02:12,864][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:02:13,363][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:02:13,865][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:02:14,365][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:02:14,864][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:02:15,363][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:02:15,865][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:02:16,364][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:02:16,863][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:02:17,363][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:02:17,861][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:02:18,362][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:02:18,860][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:02:19,358][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:02:19,856][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:02:20,354][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:02:20,852][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:02:21,350][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:02:21,851][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:02:22,350][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:02:22,849][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:02:23,349][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:02:23,848][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:02:24,347][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:02:24,847][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:02:25,347][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:02:25,845][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:02:26,342][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:02:26,838][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:02:27,334][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:02:27,831][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:02:28,329][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:02:28,826][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:02:29,324][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:02:29,823][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:02:30,321][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:02:30,820][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:02:31,318][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:02:31,815][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:02:32,313][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:02:32,811][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:02:33,310][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:02:33,808][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:02:34,308][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:02:34,806][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:02:35,304][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:02:35,802][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:02:36,301][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:02:36,800][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:02:37,298][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:02:37,797][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:02:38,296][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:02:38,795][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:02:39,293][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:02:39,790][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:02:40,288][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:02:40,786][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:02:41,284][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:02:41,783][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:02:42,282][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:02:42,781][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:02:43,280][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:02:43,778][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:02:44,275][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:02:44,774][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:02:45,274][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:02:45,776][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:02:46,276][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:02:46,775][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:02:47,274][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:02:47,772][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:02:48,274][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:02:48,773][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:02:49,273][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:02:49,771][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:02:50,270][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:02:50,768][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:02:51,267][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:02:51,765][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:02:52,264][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:02:52,763][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:02:53,262][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:02:53,762][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:02:54,261][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:02:54,761][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:02:55,261][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:02:55,760][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:02:56,259][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:02:56,759][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:02:57,258][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:02:57,756][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:02:58,253][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:02:58,750][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:02:59,247][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:02:59,745][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:03:00,243][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:03:00,741][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:03:01,239][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:03:01,737][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:03:02,235][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:03:02,733][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:03:03,231][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:03:03,729][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:03:04,228][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:03:04,727][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:03:05,225][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:03:05,724][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:03:06,223][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:03:06,722][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:03:07,220][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:03:07,718][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:03:08,216][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:03:08,714][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:03:09,212][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:03:09,710][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:03:10,209][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:03:10,707][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:03:11,205][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:03:11,703][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:03:12,200][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:03:12,698][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:03:13,197][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 01:03:13,832][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 01:03:14,563][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:03:14,565][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:03:14,567][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:03:15,213][__main__][INFO] - Iteration 208 took 1m 14s (9.03% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 12m 33s. Estimated total time: 61h 41m 6s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 22s, 500 more iterations: 10h 16m 51s. [2026-03-26 01:03:15,216][__main__][INFO] - Starting iteration 208. [2026-03-26 01:03:15,616][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:03:15,617][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:03:21,969][__main__][INFO] - Number of regex retries in iteration 208: 0 [2026-03-26 01:03:22,250][__main__][INFO] - agents played in iteration 208 are Alice, Bob [2026-03-26 01:03:23,124][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:03:23,169][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:03:23,169][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:03:23,170][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:03:23,772][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:03:24,225][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:03:24,728][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:03:25,227][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:03:25,726][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:03:26,224][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:03:26,723][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:03:27,222][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:03:27,720][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:03:28,219][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:03:28,718][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:03:29,216][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:03:29,714][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:03:30,213][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:03:30,711][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:03:31,210][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:03:31,709][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:03:32,208][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:03:32,706][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:03:33,205][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:03:33,704][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:03:34,203][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:03:34,702][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:03:35,201][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:03:35,699][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:03:36,197][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:03:36,695][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:03:37,193][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:03:37,691][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:03:38,190][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:03:38,691][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:03:39,190][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:03:39,690][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:03:40,189][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:03:40,687][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:03:41,186][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:03:41,685][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:03:42,184][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:03:42,683][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:03:43,181][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:03:43,680][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:03:44,179][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:03:44,678][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:03:45,176][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:03:45,675][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:03:46,172][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:03:46,672][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:03:47,171][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:03:47,670][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:03:48,169][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:03:48,669][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:03:49,168][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:03:49,666][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:03:50,165][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:03:50,664][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:03:51,161][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:03:51,660][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:03:52,157][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:03:52,655][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:03:53,152][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:03:53,650][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:03:54,149][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:03:54,646][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:03:55,145][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:03:55,644][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:03:56,141][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:03:56,640][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:03:57,138][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:03:57,635][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:03:58,133][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:03:58,630][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:03:59,128][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:03:59,627][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:04:00,125][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:04:00,623][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:04:01,121][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:04:01,619][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:04:02,117][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:04:02,615][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:04:03,113][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:04:03,612][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:04:04,110][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:04:04,609][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:04:05,106][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:04:05,604][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:04:06,101][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:04:06,600][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:04:07,099][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:04:07,596][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:04:08,095][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:04:08,593][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:04:09,091][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:04:09,589][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:04:10,088][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:04:10,587][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:04:11,085][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:04:11,584][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:04:12,083][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:04:12,582][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:04:13,079][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:04:13,577][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:04:14,074][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:04:14,572][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:04:15,070][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:04:15,572][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:04:16,074][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:04:16,574][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:04:17,072][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:04:17,570][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:04:18,070][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:04:18,572][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:04:19,072][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:04:19,571][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:04:20,070][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:04:20,568][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:04:21,067][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:04:21,566][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:04:22,065][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:04:22,565][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:04:23,065][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:04:23,564][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:04:24,063][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:04:24,560][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:04:25,058][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:04:25,556][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:04:26,053][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:04:26,551][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:04:27,050][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:04:27,547][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 01:04:28,178][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:04:28,916][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:04:28,918][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:04:28,920][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:04:29,762][__main__][INFO] - Iteration 209 took 1m 14s (8.95% Gen, 89.92% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 17m 33s. Estimated total time: 61h 47m 20s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 34s, 500 more iterations: 10h 17m 53s. [2026-03-26 01:04:29,764][__main__][INFO] - Starting iteration 209. [2026-03-26 01:04:30,168][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:04:30,168][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:04:36,608][__main__][INFO] - Number of regex retries in iteration 209: 0 [2026-03-26 01:04:36,608][__main__][INFO] - agents played in iteration 209 are Alice, Bob [2026-03-26 01:04:37,482][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:04:37,526][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:04:37,527][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:04:37,528][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:04:38,134][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:04:38,589][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:04:39,093][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:04:39,592][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:04:40,090][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:04:40,589][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:04:41,088][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:04:41,588][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:04:42,086][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:04:42,588][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:04:43,088][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:04:43,587][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:04:44,086][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:04:44,585][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:04:45,086][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:04:45,586][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:04:46,085][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:04:46,585][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:04:47,084][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:04:47,584][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:04:48,086][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:04:48,587][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:04:49,086][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:04:49,585][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:04:50,084][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:04:50,585][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:04:51,085][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:04:51,586][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:04:52,086][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:04:52,588][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:04:53,088][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:04:53,588][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:04:54,088][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:04:54,587][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:04:55,087][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:04:55,586][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:04:56,085][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:04:56,586][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:04:57,085][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:04:57,584][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:04:58,083][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:04:58,581][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:04:59,079][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:04:59,577][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:05:00,075][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:05:00,573][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:05:01,073][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:05:01,571][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:05:02,070][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:05:02,568][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:05:03,068][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:05:03,567][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:05:04,066][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:05:04,566][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:05:05,066][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:05:05,565][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:05:06,063][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:05:06,561][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:05:07,060][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:05:07,557][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:05:08,054][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:05:08,555][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:05:09,053][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:05:09,551][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:05:10,049][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:05:10,547][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:05:11,047][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:05:11,545][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:05:12,043][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:05:12,542][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:05:13,041][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:05:13,539][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:05:14,037][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:05:14,536][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:05:15,034][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:05:15,533][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:05:16,031][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:05:16,530][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:05:17,030][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:05:17,528][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:05:18,028][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:05:18,528][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:05:19,026][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:05:19,525][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:05:20,022][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:05:20,520][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:05:21,018][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:05:21,516][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:05:22,014][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:05:22,514][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:05:23,015][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:05:23,515][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:05:24,017][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:05:24,517][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:05:25,017][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:05:25,518][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:05:26,018][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:05:26,517][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:05:27,015][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:05:27,514][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:05:28,013][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:05:28,512][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:05:29,010][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:05:29,509][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:05:30,008][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:05:30,507][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:05:31,004][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:05:31,501][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:05:31,999][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:05:32,497][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:05:32,995][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:05:33,493][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:05:33,992][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:05:34,490][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:05:34,989][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:05:35,488][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:05:35,987][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:05:36,487][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:05:36,986][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:05:37,485][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:05:37,984][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:05:38,481][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:05:38,979][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:05:39,476][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:05:39,973][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:05:40,472][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:05:40,972][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:05:41,470][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:05:41,968][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 01:05:42,616][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 01:05:43,354][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:05:43,356][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:05:43,358][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:05:44,010][__main__][INFO] - Iteration 210 took 1m 13s (8.72% Gen, 90.39% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 1m 8s. Estimated total time: 61h 32m 10s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 21s. [2026-03-26 01:05:44,013][__main__][INFO] - Starting iteration 210. [2026-03-26 01:05:44,418][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:05:44,419][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:05:47,302][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:05:47,958][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:05:51,015][__main__][INFO] - Number of regex retries in iteration 210: 2 [2026-03-26 01:05:51,016][__main__][INFO] - agents played in iteration 210 are Alice, Bob [2026-03-26 01:05:51,892][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:05:51,937][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:05:51,937][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:05:51,938][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:05:52,557][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:05:53,012][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:05:53,516][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:05:54,015][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:05:54,514][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:05:55,014][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:05:55,513][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:05:56,013][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:05:56,512][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:05:57,011][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:05:57,509][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:05:58,008][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:05:58,506][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:05:59,004][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:05:59,502][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:06:00,001][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:06:00,499][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:06:00,997][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:06:01,496][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:06:01,994][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:06:02,492][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:06:02,990][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:06:03,488][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:06:03,987][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:06:04,488][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:06:04,986][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:06:05,483][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:06:05,981][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:06:06,478][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:06:06,975][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:06:07,472][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:06:07,969][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:06:08,467][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:06:08,968][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:06:09,469][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:06:09,969][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:06:10,468][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:06:10,969][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:06:11,469][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:06:11,969][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:06:12,470][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:06:12,970][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:06:13,470][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:06:13,969][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:06:14,469][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:06:14,969][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:06:15,468][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:06:15,968][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:06:16,469][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:06:16,968][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:06:17,466][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:06:17,965][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:06:18,466][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:06:18,965][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:06:19,466][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:06:19,965][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:06:20,465][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:06:20,966][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:06:21,466][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:06:21,967][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:06:22,467][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:06:22,969][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:06:23,469][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:06:23,969][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:06:24,468][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:06:24,968][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:06:25,467][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:06:25,966][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:06:26,466][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:06:26,965][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:06:27,464][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:06:27,964][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:06:28,462][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:06:28,960][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:06:29,458][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:06:29,955][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:06:30,454][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:06:30,952][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:06:31,451][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:06:31,949][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:06:32,448][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:06:32,946][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:06:33,444][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:06:33,941][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:06:34,439][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:06:34,936][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:06:35,434][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:06:35,932][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:06:36,431][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:06:36,929][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:06:37,426][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:06:37,923][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:06:38,420][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:06:38,917][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:06:39,414][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:06:39,912][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:06:40,410][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:06:40,910][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:06:41,409][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:06:41,910][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:06:42,411][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:06:42,911][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:06:43,412][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:06:43,912][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:06:44,412][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:06:44,912][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:06:45,411][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:06:45,910][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:06:46,410][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:06:46,909][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:06:47,409][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:06:47,909][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:06:48,410][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:06:48,909][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:06:49,408][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:06:49,907][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:06:50,407][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:06:50,906][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:06:51,406][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:06:51,904][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:06:52,402][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:06:52,902][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:06:53,401][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:06:53,901][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:06:54,401][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:06:54,901][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:06:55,402][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:06:55,901][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:06:56,401][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21724 tokens. [2026-03-26 01:06:57,051][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:06:57,789][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:06:57,791][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:06:57,793][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:06:58,452][__main__][INFO] - Iteration 211 took 1m 14s (8.91% Gen, 90.20% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 9m 28s. Estimated total time: 61h 41m 44s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 23s, 500 more iterations: 10h 16m 57s. [2026-03-26 01:06:58,455][__main__][INFO] - Starting iteration 211. [2026-03-26 01:06:58,854][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:06:58,855][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:07:04,913][__main__][INFO] - Number of regex retries in iteration 211: 0 [2026-03-26 01:07:04,914][__main__][INFO] - agents played in iteration 211 are Alice, Bob [2026-03-26 01:07:05,787][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:07:06,117][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:07:06,118][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:07:06,119][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:07:06,735][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:07:07,189][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:07:07,693][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:07:08,193][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:07:08,692][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:07:09,192][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:07:09,691][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:07:10,190][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:07:10,689][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:07:11,190][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:07:11,689][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:07:12,188][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:07:12,687][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:07:13,186][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:07:13,685][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:07:14,184][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:07:14,683][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:07:15,182][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:07:15,680][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:07:16,178][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:07:16,677][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:07:17,174][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:07:17,671][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:07:18,170][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:07:18,670][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:07:19,169][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:07:19,669][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:07:20,169][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:07:20,669][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:07:21,167][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:07:21,666][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:07:22,165][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:07:22,665][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:07:23,164][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:07:23,666][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:07:24,166][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:07:24,665][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:07:25,164][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:07:25,663][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:07:26,161][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:07:26,661][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:07:27,160][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:07:27,657][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:07:28,156][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:07:28,654][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:07:29,152][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:07:29,650][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:07:30,148][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:07:30,648][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:07:31,150][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:07:31,650][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:07:32,150][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:07:32,650][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:07:33,150][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:07:33,650][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:07:34,151][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:07:34,653][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:07:35,154][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:07:35,656][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:07:36,156][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:07:36,655][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:07:37,154][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:07:37,653][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:07:38,153][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:07:38,653][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:07:39,152][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:07:39,650][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:07:40,149][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:07:40,647][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:07:41,145][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:07:41,646][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:07:42,145][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:07:42,644][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:07:43,145][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:07:43,643][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:07:44,142][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:07:44,640][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:07:45,139][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:07:45,638][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:07:46,137][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:07:46,636][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:07:47,135][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:07:47,637][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:07:48,136][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:07:48,635][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:07:49,134][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:07:49,632][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:07:50,131][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:07:50,630][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:07:51,128][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:07:51,628][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:07:52,128][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:07:52,628][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:07:53,128][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:07:53,628][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:07:54,127][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:07:54,626][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:07:55,126][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:07:55,626][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:07:56,126][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:07:56,628][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:07:57,127][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:07:57,626][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:07:58,126][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:07:58,625][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:07:59,124][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:07:59,622][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:08:00,119][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:08:00,617][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:08:01,115][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:08:01,614][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:08:02,112][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:08:02,611][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:08:03,111][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:08:03,612][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:08:04,113][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:08:04,614][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:08:05,115][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:08:05,616][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:08:06,116][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:08:06,616][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:08:07,116][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:08:07,616][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:08:08,115][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:08:08,616][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:08:09,115][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:08:09,615][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:08:10,115][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:08:10,614][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:08:11,280][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 01:08:12,022][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:08:12,024][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:08:12,026][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:08:12,685][__main__][INFO] - Iteration 212 took 1m 13s (8.21% Gen, 90.90% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 58m 2s. Estimated total time: 61h 31m 32s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 3s, 500 more iterations: 10h 15m 15s. [2026-03-26 01:08:12,687][__main__][INFO] - Starting iteration 212. [2026-03-26 01:08:13,088][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:08:13,088][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:08:16,322][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:08:19,615][__main__][INFO] - Number of regex retries in iteration 212: 1 [2026-03-26 01:08:19,616][__main__][INFO] - agents played in iteration 212 are Alice, Bob [2026-03-26 01:08:20,494][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:08:20,538][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:08:20,539][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:08:20,540][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:08:21,157][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:08:21,613][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:08:22,117][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:08:22,618][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:08:23,117][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:08:23,616][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:08:24,114][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:08:24,613][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:08:25,113][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:08:25,614][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:08:26,114][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:08:26,614][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:08:27,114][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:08:27,614][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:08:28,114][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:08:28,613][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:08:29,113][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:08:29,612][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:08:30,118][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:08:30,619][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:08:31,120][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:08:31,620][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:08:32,120][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:08:32,620][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:08:33,120][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:08:33,620][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:08:34,119][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:08:34,618][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:08:35,117][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:08:35,616][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:08:36,123][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:08:36,626][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:08:37,127][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:08:37,629][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:08:38,130][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:08:38,631][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:08:39,131][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:08:39,631][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:08:40,131][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:08:40,631][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:08:41,131][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:08:41,631][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:08:42,138][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:08:42,640][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:08:43,141][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:08:43,642][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:08:44,143][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:08:44,645][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:08:45,147][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:08:45,649][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:08:46,150][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:08:46,650][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:08:47,149][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:08:47,649][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:08:48,162][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:08:48,663][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:08:49,162][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:08:49,663][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:08:50,164][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:08:50,666][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:08:51,167][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:08:51,668][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:08:52,168][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:08:52,669][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:08:53,170][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:08:53,670][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:08:54,171][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:08:54,671][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:08:55,172][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:08:55,671][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:08:56,171][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:08:56,671][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:08:57,171][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:08:57,671][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:08:58,173][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:08:58,674][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:08:59,174][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:08:59,673][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:09:00,175][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:09:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:09:01,175][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:09:01,675][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:09:02,175][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:09:02,675][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:09:03,173][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:09:03,673][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:09:04,174][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:09:04,673][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:09:05,174][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:09:05,673][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:09:06,172][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:09:06,672][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:09:07,171][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:09:07,670][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:09:08,170][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:09:08,669][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:09:09,168][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:09:09,667][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:09:10,165][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:09:10,664][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:09:11,162][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:09:11,666][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:09:12,166][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:09:12,666][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:09:13,165][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:09:13,664][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:09:14,162][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:09:14,661][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:09:15,160][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:09:15,658][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:09:16,157][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:09:16,655][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:09:17,154][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:09:17,652][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:09:18,152][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:09:18,651][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:09:19,150][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:09:19,649][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:09:20,148][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:09:20,646][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:09:21,144][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:09:21,643][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:09:22,143][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:09:22,642][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:09:23,142][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:09:23,642][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:09:24,142][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:09:24,642][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:09:25,142][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 01:09:25,800][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:09:26,563][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:09:26,566][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:09:26,567][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:09:27,270][__main__][INFO] - Iteration 213 took 1m 14s (8.80% Gen, 90.25% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 14m 24s. Estimated total time: 61h 49m 9s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 38s, 500 more iterations: 10h 18m 11s. [2026-03-26 01:09:27,272][__main__][INFO] - Starting iteration 213. [2026-03-26 01:09:27,673][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:09:27,674][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:09:33,734][__main__][INFO] - Number of regex retries in iteration 213: 0 [2026-03-26 01:09:33,735][__main__][INFO] - agents played in iteration 213 are Alice, Bob [2026-03-26 01:09:34,620][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:09:34,665][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:09:34,665][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:09:34,666][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:09:35,288][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:09:35,744][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:09:36,247][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:09:36,747][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:09:37,247][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:09:37,746][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:09:38,244][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:09:38,742][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:09:39,241][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:09:39,739][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:09:40,237][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:09:40,736][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:09:41,234][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:09:41,732][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:09:42,230][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:09:42,729][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:09:43,227][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:09:43,726][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:09:44,224][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:09:44,722][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:09:45,219][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:09:45,717][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:09:46,216][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:09:46,713][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:09:47,211][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:09:47,711][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:09:48,212][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:09:48,711][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:09:49,210][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:09:49,709][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:09:50,207][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:09:50,704][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:09:51,202][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:09:51,700][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:09:52,199][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:09:52,698][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:09:53,197][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:09:53,695][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:09:54,193][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:09:54,692][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:09:55,191][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:09:55,690][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:09:56,189][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:09:56,687][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:09:57,187][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:09:57,685][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:09:58,183][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:09:58,681][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:09:59,179][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:09:59,679][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:10:00,179][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:10:00,678][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:10:01,178][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:10:01,677][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:10:02,177][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:10:02,676][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:10:03,175][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:10:03,675][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:10:04,176][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:10:04,676][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:10:05,175][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:10:05,674][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:10:06,173][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:10:06,673][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:10:07,174][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:10:07,674][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:10:08,174][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:10:08,672][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:10:09,171][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:10:09,670][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:10:10,169][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:10:10,668][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:10:11,167][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:10:11,670][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:10:12,171][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:10:12,671][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:10:13,172][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:10:13,672][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:10:14,173][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:10:14,674][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:10:15,176][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:10:15,678][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:10:16,177][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:10:16,674][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:10:17,180][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:10:17,680][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:10:18,181][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:10:18,681][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:10:19,180][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:10:19,681][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:10:20,180][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:10:20,680][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:10:21,180][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:10:21,680][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:10:22,180][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:10:22,680][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:10:23,189][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:10:23,691][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:10:24,191][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:10:24,692][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:10:25,193][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:10:25,693][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:10:26,192][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:10:26,693][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:10:27,193][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:10:27,694][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:10:28,195][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:10:28,695][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:10:29,205][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:10:29,706][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:10:30,207][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:10:30,707][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:10:31,208][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:10:31,708][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:10:32,209][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:10:32,709][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:10:33,209][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:10:33,710][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:10:34,210][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:10:34,711][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:10:35,216][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:10:35,715][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:10:36,215][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:10:36,715][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:10:37,215][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:10:37,715][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:10:38,214][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:10:38,714][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:10:39,213][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21747 tokens. [2026-03-26 01:10:39,856][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 01:10:40,622][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:10:40,625][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:10:40,626][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:10:41,311][__main__][INFO] - Iteration 214 took 1m 13s (8.23% Gen, 90.84% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 45m 55s. Estimated total time: 61h 21m 54s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 43s, 500 more iterations: 10h 13m 39s. [2026-03-26 01:10:41,313][__main__][INFO] - Starting iteration 214. [2026-03-26 01:10:41,716][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:10:41,717][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:10:48,081][__main__][INFO] - Number of regex retries in iteration 214: 0 [2026-03-26 01:10:48,081][__main__][INFO] - agents played in iteration 214 are Alice, Bob [2026-03-26 01:10:48,959][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:10:49,004][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:10:49,004][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:10:49,005][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:10:49,610][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:10:50,067][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:10:50,570][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:10:51,070][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:10:51,570][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:10:52,070][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:10:52,569][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:10:53,069][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:10:53,569][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:10:54,068][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:10:54,569][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:10:55,070][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:10:55,569][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:10:56,068][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:10:56,567][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:10:57,066][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:10:57,565][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:10:58,066][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:10:58,566][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:10:59,066][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:10:59,564][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:11:00,062][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:11:00,562][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:11:01,060][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:11:01,559][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:11:02,056][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:11:02,555][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:11:03,053][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:11:03,551][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:11:04,049][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:11:04,547][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:11:05,045][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:11:05,543][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:11:06,042][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:11:06,542][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:11:07,041][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:11:07,540][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:11:08,040][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:11:08,539][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:11:09,039][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:11:09,538][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:11:10,037][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:11:10,537][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:11:11,036][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:11:11,535][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:11:12,034][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:11:12,534][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:11:13,033][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:11:13,532][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:11:14,030][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:11:14,528][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:11:15,027][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:11:15,524][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:11:16,022][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:11:16,521][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:11:17,018][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:11:17,516][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:11:18,014][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:11:18,515][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:11:19,014][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:11:19,513][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:11:20,010][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:11:20,508][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:11:21,005][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:11:21,503][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:11:22,003][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:11:22,504][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:11:23,004][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:11:23,504][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:11:24,003][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:11:24,503][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:11:25,002][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:11:25,503][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:11:26,002][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:11:26,502][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:11:27,002][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:11:27,501][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:11:27,999][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:11:28,499][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:11:28,998][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:11:29,496][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:11:29,995][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:11:30,494][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:11:30,993][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:11:31,492][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:11:31,991][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:11:32,491][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:11:32,990][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:11:33,488][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:11:33,987][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:11:34,486][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:11:34,984][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:11:35,481][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:11:35,978][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:11:36,475][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:11:36,973][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:11:37,471][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:11:37,970][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:11:38,470][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:11:38,970][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:11:39,469][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:11:39,968][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:11:40,468][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:11:40,967][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:11:41,469][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:11:41,968][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:11:42,467][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:11:42,965][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:11:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:11:43,962][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:11:44,461][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:11:44,960][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:11:45,461][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:11:45,961][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:11:46,459][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:11:46,957][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:11:47,454][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:11:47,953][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:11:48,452][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:11:48,951][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:11:49,450][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:11:49,948][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:11:50,446][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:11:50,944][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:11:51,442][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:11:51,940][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:11:52,438][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:11:52,935][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:11:53,433][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:11:54,080][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:11:54,959][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:11:54,962][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:11:54,974][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:11:55,839][__main__][INFO] - Iteration 215 took 1m 14s (8.59% Gen, 90.24% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 8m 58s. Estimated total time: 61h 46m 11s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 32s, 500 more iterations: 10h 17m 41s. [2026-03-26 01:11:55,841][__main__][INFO] - Starting iteration 215. [2026-03-26 01:11:56,239][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:11:56,240][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:11:57,343][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:11:57,900][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:12:02,386][__main__][INFO] - Number of regex retries in iteration 215: 2 [2026-03-26 01:12:02,386][__main__][INFO] - agents played in iteration 215 are Alice, Bob [2026-03-26 01:12:03,265][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:12:03,590][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:12:03,591][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:12:03,591][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:12:04,194][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:12:04,650][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:12:05,152][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:12:05,651][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:12:06,151][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:12:06,651][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:12:07,150][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:12:07,649][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:12:08,148][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:12:08,646][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:12:09,144][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:12:09,641][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:12:10,139][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:12:10,637][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:12:11,135][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:12:11,633][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:12:12,131][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:12:12,629][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:12:13,127][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:12:13,625][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:12:14,123][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:12:14,620][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:12:15,117][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:12:15,614][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:12:16,111][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:12:16,611][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:12:17,110][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:12:17,608][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:12:18,108][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:12:18,608][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:12:19,105][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:12:19,603][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:12:20,101][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:12:20,602][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:12:21,102][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:12:21,601][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:12:22,101][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:12:22,601][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:12:23,100][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:12:23,599][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:12:24,099][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:12:24,598][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:12:25,097][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:12:25,595][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:12:26,094][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:12:26,593][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:12:27,092][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:12:27,591][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:12:28,091][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:12:28,593][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:12:29,094][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:12:29,593][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:12:30,094][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:12:30,594][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:12:31,094][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:12:31,595][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:12:32,095][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:12:32,594][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:12:33,091][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:12:33,589][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:12:34,087][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:12:34,583][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:12:35,081][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:12:35,578][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:12:36,075][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:12:36,574][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:12:37,073][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:12:37,570][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:12:38,070][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:12:38,569][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:12:39,067][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:12:39,565][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:12:40,065][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:12:40,566][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:12:41,063][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:12:41,560][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:12:42,058][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:12:42,555][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:12:43,052][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:12:43,550][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:12:44,050][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:12:44,547][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:12:45,044][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:12:45,541][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:12:46,038][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:12:46,534][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:12:47,031][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:12:47,527][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:12:48,026][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:12:48,526][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:12:49,024][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:12:49,521][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:12:50,019][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:12:50,517][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:12:51,015][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:12:51,513][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:12:52,011][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:12:52,511][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:12:53,011][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:12:53,510][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:12:54,010][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:12:54,510][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:12:55,010][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:12:55,509][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:12:56,009][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:12:56,508][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:12:57,007][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:12:57,506][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:12:58,006][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:12:58,504][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:12:59,003][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:12:59,501][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:12:59,999][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:13:00,498][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:13:00,998][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:13:01,497][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:13:01,997][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:13:02,497][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:13:02,997][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:13:03,496][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:13:03,996][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:13:04,493][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:13:04,992][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:13:05,490][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:13:05,985][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:13:06,484][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:13:06,979][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:13:07,477][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:13:07,976][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21673 tokens. [2026-03-26 01:13:08,625][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 01:13:09,354][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:13:09,357][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:13:09,358][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:13:10,082][__main__][INFO] - Iteration 216 took 1m 13s (8.32% Gen, 90.69% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 53m 42s. Estimated total time: 61h 32m 10s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 21s. [2026-03-26 01:13:10,084][__main__][INFO] - Starting iteration 216. [2026-03-26 01:13:10,487][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:13:10,487][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:13:16,573][__main__][INFO] - Number of regex retries in iteration 216: 0 [2026-03-26 01:13:16,574][__main__][INFO] - agents played in iteration 216 are Alice, Bob [2026-03-26 01:13:17,453][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:13:17,498][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:13:17,498][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:13:17,499][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:13:18,109][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:13:18,563][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:13:19,066][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:13:19,565][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:13:20,063][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:13:20,560][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:13:21,058][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:13:21,556][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:13:22,054][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:13:22,553][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:13:23,051][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:13:23,550][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:13:24,049][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:13:24,547][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:13:25,045][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:13:25,543][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:13:26,041][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:13:26,539][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:13:27,037][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:13:27,536][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:13:28,035][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:13:28,534][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:13:29,033][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:13:29,532][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:13:30,030][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:13:30,531][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:13:31,031][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:13:31,530][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:13:32,029][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:13:32,528][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:13:33,026][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:13:33,525][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:13:34,023][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:13:34,521][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:13:35,017][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:13:35,515][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:13:36,014][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:13:36,511][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:13:37,008][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:13:37,505][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:13:38,002][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:13:38,500][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:13:39,000][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:13:39,499][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:13:39,998][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:13:40,496][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:13:40,995][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:13:41,495][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:13:41,994][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:13:42,492][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:13:42,991][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:13:43,489][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:13:43,987][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:13:44,485][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:13:44,982][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:13:45,480][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:13:45,977][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:13:46,477][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:13:46,977][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:13:47,478][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:13:47,979][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:13:48,480][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:13:48,980][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:13:49,479][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:13:49,978][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:13:50,476][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:13:50,974][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:13:51,473][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:13:51,971][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:13:52,470][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:13:52,968][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:13:53,467][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:13:53,965][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:13:54,463][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:13:54,961][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:13:55,459][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:13:55,958][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:13:56,456][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:13:56,954][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:13:57,452][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:13:57,950][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:13:58,448][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:13:58,948][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:13:59,447][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:13:59,946][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:14:00,444][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:14:00,942][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:14:01,441][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:14:01,939][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:14:02,436][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:14:02,934][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:14:03,431][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:14:03,929][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:14:04,427][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:14:04,925][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:14:05,422][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:14:05,920][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:14:06,417][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:14:06,915][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:14:07,412][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:14:07,911][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:14:08,408][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:14:08,906][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:14:09,403][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:14:09,900][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:14:10,399][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:14:10,897][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:14:11,395][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:14:11,895][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:14:12,395][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:14:12,894][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:14:13,392][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:14:13,892][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:14:14,390][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:14:14,888][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:14:15,387][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:14:15,885][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:14:16,383][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:14:16,880][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:14:17,377][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:14:17,875][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:14:18,377][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:14:18,877][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:14:19,376][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:14:19,876][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:14:20,377][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:14:20,877][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:14:21,376][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:14:21,876][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 01:14:22,527][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:14:23,262][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:14:23,264][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:14:23,266][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:14:23,913][__main__][INFO] - Iteration 217 took 1m 13s (8.29% Gen, 90.83% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 31m 39s. Estimated total time: 61h 11m 21s. Time estimates for 10 more iterations: 12m 14s, 100 more iterations: 2h 2m 22s, 500 more iterations: 10h 11m 53s. [2026-03-26 01:14:23,915][__main__][INFO] - Starting iteration 217. [2026-03-26 01:14:24,319][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:14:24,320][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:14:31,261][__main__][INFO] - Number of regex retries in iteration 217: 0 [2026-03-26 01:14:31,262][__main__][INFO] - agents played in iteration 217 are Alice, Bob [2026-03-26 01:14:32,148][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:14:32,192][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:14:32,193][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:14:32,194][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:14:32,795][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:14:33,252][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:14:33,756][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:14:34,256][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:14:34,755][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:14:35,255][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:14:35,754][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:14:36,254][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:14:36,754][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:14:37,252][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:14:37,752][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:14:38,251][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:14:38,757][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:14:39,258][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:14:39,759][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:14:40,259][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:14:40,760][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:14:41,259][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:14:41,757][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:14:42,256][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:14:42,754][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:14:43,250][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:14:43,751][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:14:44,258][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:14:44,759][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:14:45,260][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:14:45,763][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:14:46,265][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:14:46,766][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:14:47,266][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:14:47,766][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:14:48,267][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:14:48,767][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:14:49,269][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:14:49,771][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:14:50,277][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:14:50,779][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:14:51,281][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:14:51,782][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:14:52,283][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:14:52,786][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:14:53,287][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:14:53,787][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:14:54,287][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:14:54,787][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:14:55,288][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:14:55,792][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:14:56,292][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:14:56,792][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:14:57,291][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:14:57,791][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:14:58,290][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:14:58,790][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:14:59,290][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:14:59,789][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:15:00,288][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:15:00,788][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:15:01,289][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:15:01,790][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:15:02,290][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:15:02,790][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:15:03,290][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:15:03,790][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:15:04,290][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:15:04,790][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:15:05,289][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:15:05,790][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:15:06,289][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:15:06,788][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:15:07,287][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:15:07,786][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:15:08,284][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:15:08,782][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:15:09,279][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:15:09,777][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:15:10,275][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:15:10,775][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:15:11,274][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:15:11,773][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:15:12,272][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:15:12,770][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:15:13,268][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:15:13,766][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:15:14,262][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:15:14,760][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:15:15,258][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:15:15,755][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:15:16,253][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:15:16,750][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:15:17,249][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:15:17,747][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:15:18,246][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:15:18,746][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:15:19,244][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:15:19,742][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:15:20,241][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:15:20,738][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:15:21,238][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:15:21,737][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:15:22,237][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:15:22,737][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:15:23,237][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:15:23,737][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:15:24,237][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:15:24,737][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:15:25,237][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:15:25,736][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:15:26,236][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:15:26,735][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:15:27,234][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:15:27,733][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:15:28,232][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:15:28,731][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:15:29,230][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:15:29,728][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:15:30,226][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:15:30,725][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:15:31,223][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:15:31,722][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:15:32,219][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:15:32,717][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:15:33,217][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:15:33,716][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:15:34,216][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:15:34,716][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:15:35,216][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:15:35,716][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:15:36,215][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:15:36,714][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21732 tokens. [2026-03-26 01:15:37,359][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.31%, Current % of VRAM taken: 60.67%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 01:15:38,091][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:15:38,094][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:15:38,095][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:15:38,810][__main__][INFO] - Iteration 218 took 1m 14s (9.32% Gen, 89.72% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 23m 39s. Estimated total time: 62h 4m 35s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 45s. [2026-03-26 01:15:38,813][__main__][INFO] - Starting iteration 218. [2026-03-26 01:15:39,212][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:15:39,213][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:15:39,810][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:15:44,395][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:15:46,057][__main__][INFO] - Number of regex retries in iteration 218: 2 [2026-03-26 01:15:46,058][__main__][INFO] - agents played in iteration 218 are Alice, Bob [2026-03-26 01:15:46,941][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:15:46,986][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:15:46,986][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:15:46,987][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:15:47,588][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:15:48,044][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:15:48,548][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:15:49,046][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:15:49,544][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:15:50,042][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:15:50,539][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:15:51,038][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:15:51,536][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:15:52,036][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:15:52,534][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:15:53,032][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:15:53,530][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:15:54,031][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:15:54,529][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:15:55,028][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:15:55,528][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:15:56,026][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:15:56,524][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:15:57,022][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:15:57,521][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:15:58,023][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:15:58,521][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:15:59,020][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:15:59,519][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:16:00,017][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:16:00,515][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:16:01,012][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:16:01,510][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:16:02,009][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:16:02,507][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:16:03,006][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:16:03,505][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:16:04,002][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:16:04,500][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:16:04,997][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:16:05,495][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:16:05,994][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:16:06,493][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:16:06,991][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:16:07,490][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:16:07,989][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:16:08,487][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:16:08,985][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:16:09,483][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:16:09,980][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:16:10,477][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:16:10,976][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:16:11,474][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:16:11,974][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:16:12,473][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:16:12,972][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:16:13,472][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:16:13,971][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:16:14,470][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:16:14,970][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:16:15,471][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:16:15,972][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:16:16,469][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:16:16,967][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:16:17,464][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:16:17,961][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:16:18,461][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:16:18,958][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:16:19,457][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:16:19,955][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:16:20,453][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:16:20,951][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:16:21,450][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:16:21,948][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:16:22,448][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:16:22,948][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:16:23,446][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:16:23,945][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:16:24,443][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:16:24,941][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:16:25,440][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:16:25,938][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:16:26,437][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:16:26,936][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:16:27,435][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:16:27,933][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:16:28,431][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:16:28,929][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:16:29,427][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:16:29,924][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:16:30,422][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:16:30,920][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:16:31,419][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:16:31,917][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:16:32,415][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:16:32,913][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:16:33,411][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:16:33,909][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:16:34,406][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:16:34,904][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:16:35,401][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:16:35,899][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:16:36,397][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:16:36,895][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:16:37,392][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:16:37,891][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:16:38,389][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:16:38,887][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:16:39,385][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:16:39,885][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:16:40,383][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:16:40,882][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:16:41,380][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:16:41,879][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:16:42,377][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:16:42,876][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:16:43,375][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:16:43,875][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:16:44,374][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:16:44,873][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:16:45,372][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:16:45,870][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:16:46,369][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:16:46,869][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:16:47,370][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:16:47,868][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:16:48,368][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:16:48,866][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:16:49,363][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:16:49,860][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:16:50,357][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:16:50,854][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:16:51,352][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 01:16:51,996][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:16:52,728][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:16:52,730][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:16:52,732][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:16:53,597][__main__][INFO] - Iteration 219 took 1m 14s (9.20% Gen, 89.63% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 17m 5s. Estimated total time: 61h 59m 17s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 58s, 500 more iterations: 10h 19m 52s. [2026-03-26 01:16:53,600][__main__][INFO] - Starting iteration 219. [2026-03-26 01:16:54,004][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:16:54,005][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:17:00,493][__main__][INFO] - Number of regex retries in iteration 219: 0 [2026-03-26 01:17:00,494][__main__][INFO] - agents played in iteration 219 are Alice, Bob [2026-03-26 01:17:01,367][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:17:01,412][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:17:01,412][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:17:01,413][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:17:02,019][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:17:02,476][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:17:02,982][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:17:03,481][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:17:03,980][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:17:04,479][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:17:04,978][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:17:05,478][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:17:05,976][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:17:06,478][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:17:06,979][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:17:07,479][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:17:07,979][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:17:08,478][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:17:08,978][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:17:09,478][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:17:09,978][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:17:10,477][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:17:10,977][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:17:11,476][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:17:11,975][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:17:12,473][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:17:12,973][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:17:13,471][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:17:13,971][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:17:14,471][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:17:14,970][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:17:15,469][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:17:15,968][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:17:16,466][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:17:16,964][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:17:17,463][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:17:17,961][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:17:18,463][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:17:18,963][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:17:19,462][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:17:19,961][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:17:20,460][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:17:20,958][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:17:21,458][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:17:21,958][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:17:22,456][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:17:22,954][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:17:23,451][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:17:23,949][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:17:24,447][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:17:24,945][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:17:25,443][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:17:25,940][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:17:26,438][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:17:26,936][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:17:27,434][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:17:27,932][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:17:28,430][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:17:28,927][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:17:29,424][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:17:29,921][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:17:30,420][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:17:30,918][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:17:31,416][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:17:31,914][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:17:32,413][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:17:32,911][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:17:33,408][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:17:33,906][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:17:34,405][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:17:34,905][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:17:35,404][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:17:35,903][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:17:36,403][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:17:36,901][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:17:37,400][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:17:37,898][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:17:38,399][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:17:38,900][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:17:39,400][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:17:39,900][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:17:40,400][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:17:40,899][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:17:41,398][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:17:41,899][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:17:42,399][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:17:42,898][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:17:43,397][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:17:43,895][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:17:44,395][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:17:44,894][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:17:45,395][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:17:45,896][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:17:46,396][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:17:46,894][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:17:47,393][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:17:47,892][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:17:48,392][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:17:48,891][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:17:49,393][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:17:49,892][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:17:50,390][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:17:50,889][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:17:51,388][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:17:51,887][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:17:52,391][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:17:52,894][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:17:53,394][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:17:53,894][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:17:54,393][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:17:54,892][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:17:55,392][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:17:55,892][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:17:56,392][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:17:56,891][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:17:57,389][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:17:57,890][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:17:58,393][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:17:58,894][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:17:59,394][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:17:59,894][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:18:00,394][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:18:00,894][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:18:01,393][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:18:01,893][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:18:02,393][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:18:02,892][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:18:03,391][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:18:03,890][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:18:04,395][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:18:04,896][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:18:05,396][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:18:05,897][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 01:18:06,543][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 01:18:07,278][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:18:07,280][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:18:07,282][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:18:08,130][__main__][INFO] - Iteration 220 took 1m 14s (8.75% Gen, 90.10% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 57h 2m 53s. Estimated total time: 61h 46m 19s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 32s, 500 more iterations: 10h 17m 43s. [2026-03-26 01:18:08,133][__main__][INFO] - Starting iteration 220. [2026-03-26 01:18:08,531][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:18:08,532][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:18:14,577][__main__][INFO] - Number of regex retries in iteration 220: 0 [2026-03-26 01:18:14,577][__main__][INFO] - agents played in iteration 220 are Alice, Bob [2026-03-26 01:18:15,450][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:18:15,494][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:18:15,495][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:18:15,496][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:18:16,105][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:18:16,564][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:18:17,069][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:18:17,569][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:18:18,069][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:18:18,571][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:18:19,072][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:18:19,571][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:18:20,071][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:18:20,571][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:18:21,071][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:18:21,572][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:18:22,072][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:18:22,572][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:18:23,072][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:18:23,572][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:18:24,071][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:18:24,569][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:18:25,070][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:18:25,570][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:18:26,070][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:18:26,571][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:18:27,071][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:18:27,570][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:18:28,070][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:18:28,569][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:18:29,068][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:18:29,567][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:18:30,065][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:18:30,563][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:18:31,062][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:18:31,559][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:18:32,058][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:18:32,558][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:18:33,057][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:18:33,556][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:18:34,055][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:18:34,554][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:18:35,054][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:18:35,552][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:18:36,052][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:18:36,550][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:18:37,050][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:18:37,548][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:18:38,047][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:18:38,545][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:18:39,043][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:18:39,541][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:18:40,039][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:18:40,539][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:18:41,038][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:18:41,538][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:18:42,038][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:18:42,537][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:18:43,037][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:18:43,537][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:18:44,036][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:18:44,534][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:18:45,031][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:18:45,530][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:18:46,029][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:18:46,527][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:18:47,024][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:18:47,522][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:18:48,021][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:18:48,522][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:18:49,022][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:18:49,521][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:18:50,020][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:18:50,519][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:18:51,019][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:18:51,519][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:18:52,018][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:18:52,519][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:18:53,018][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:18:53,517][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:18:54,016][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:18:54,516][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:18:55,016][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:18:55,516][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:18:56,017][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:18:56,516][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:18:57,015][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:18:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:18:58,014][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:18:58,514][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:18:59,012][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:18:59,511][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:19:00,009][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:19:00,509][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:19:01,007][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:19:01,505][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:19:02,003][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:19:02,502][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:19:03,001][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:19:03,498][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:19:03,996][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:19:04,497][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:19:04,997][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:19:05,496][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:19:05,995][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:19:06,495][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:19:06,994][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:19:07,494][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:19:07,993][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:19:08,492][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:19:08,991][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:19:09,490][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:19:09,989][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:19:10,487][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:19:10,985][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:19:11,482][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:19:11,981][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:19:12,481][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:19:12,981][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:19:13,481][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:19:13,980][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:19:14,479][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:19:14,977][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:19:15,479][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:19:15,980][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:19:16,479][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:19:16,977][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:19:17,474][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:19:17,972][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:19:18,471][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:19:18,969][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:19:19,465][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:19:19,962][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 01:19:20,597][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 01:19:21,331][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:19:21,333][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:19:21,335][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:19:22,155][__main__][INFO] - Iteration 221 took 1m 13s (8.21% Gen, 90.67% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 36m 33s. Estimated total time: 61h 21m 13s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 42s, 500 more iterations: 10h 13m 32s. [2026-03-26 01:19:22,157][__main__][INFO] - Starting iteration 221. [2026-03-26 01:19:22,554][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:19:22,555][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:19:23,127][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:19:23,132][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:19:29,064][__main__][INFO] - Number of regex retries in iteration 221: 2 [2026-03-26 01:19:29,065][__main__][INFO] - agents played in iteration 221 are Alice, Bob [2026-03-26 01:19:29,943][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:19:29,987][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:19:29,988][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:19:29,988][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:19:30,590][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:19:31,046][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:19:31,549][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:19:32,048][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:19:32,548][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:19:33,046][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:19:33,544][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:19:34,043][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:19:34,542][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:19:35,039][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:19:35,537][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:19:36,035][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:19:36,533][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:19:37,032][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:19:37,531][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:19:38,029][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:19:38,528][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:19:39,029][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:19:39,528][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:19:40,027][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:19:40,526][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:19:41,024][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:19:41,523][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:19:42,022][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:19:42,521][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:19:43,020][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:19:43,518][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:19:44,017][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:19:44,515][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:19:45,014][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:19:45,512][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:19:46,011][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:19:46,510][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:19:47,011][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:19:47,511][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:19:48,011][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:19:48,513][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:19:49,012][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:19:49,513][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:19:50,012][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:19:50,512][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:19:51,010][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:19:51,509][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:19:52,007][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:19:52,504][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:19:53,001][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:19:53,500][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:19:53,999][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:19:54,498][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:19:54,997][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:19:55,497][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:19:55,996][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:19:56,495][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:19:56,995][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:19:57,495][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:19:57,996][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:19:58,496][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:19:58,996][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:19:59,495][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:19:59,996][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:20:00,496][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:20:00,997][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:20:01,496][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:20:01,996][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:20:02,494][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:20:02,993][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:20:03,494][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:20:03,993][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:20:04,492][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:20:04,991][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:20:05,491][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:20:05,990][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:20:06,489][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:20:06,988][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:20:07,485][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:20:07,982][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:20:08,480][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:20:08,979][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:20:09,477][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:20:09,974][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:20:10,472][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:20:10,971][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:20:11,471][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:20:11,970][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:20:12,471][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:20:12,971][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:20:13,470][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:20:13,970][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:20:14,468][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:20:14,966][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:20:15,463][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:20:15,960][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:20:16,458][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:20:16,957][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:20:17,455][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:20:17,953][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:20:18,454][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:20:18,953][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:20:19,454][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:20:19,954][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:20:20,453][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:20:20,953][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:20:21,452][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:20:21,951][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:20:22,450][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:20:22,949][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:20:23,447][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:20:23,946][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:20:24,444][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:20:24,942][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:20:25,441][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:20:25,939][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:20:26,438][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:20:26,937][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:20:27,435][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:20:27,935][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:20:28,434][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:20:28,934][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:20:29,433][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:20:29,932][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:20:30,431][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:20:30,929][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:20:31,426][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:20:31,924][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:20:32,422][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:20:32,920][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:20:33,417][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:20:33,916][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:20:34,415][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 01:20:35,046][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 01:20:35,784][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:20:35,786][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:20:35,788][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:20:36,443][__main__][INFO] - Iteration 222 took 1m 13s (8.81% Gen, 90.30% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 48m 32s. Estimated total time: 61h 34m 27s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 8s, 500 more iterations: 10h 15m 44s. [2026-03-26 01:20:36,445][__main__][INFO] - Starting iteration 222. [2026-03-26 01:20:36,844][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:20:36,845][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:20:39,040][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:20:43,309][__main__][INFO] - Number of regex retries in iteration 222: 1 [2026-03-26 01:20:43,309][__main__][INFO] - agents played in iteration 222 are Alice, Bob [2026-03-26 01:20:44,181][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:20:44,225][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:20:44,226][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:20:44,227][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:20:44,829][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:20:45,286][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:20:45,792][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:20:46,293][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:20:46,792][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:20:47,291][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:20:47,790][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:20:48,292][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:20:48,794][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:20:49,296][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:20:49,796][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:20:50,296][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:20:50,796][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:20:51,296][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:20:51,796][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:20:52,296][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:20:52,796][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:20:53,295][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:20:53,793][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:20:54,292][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:20:54,795][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:20:55,296][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:20:55,796][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:20:56,296][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:20:56,796][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:20:57,298][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:20:57,799][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:20:58,300][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:20:58,800][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:20:59,301][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:20:59,802][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:21:00,311][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:21:00,813][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:21:01,313][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:21:01,814][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:21:02,314][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:21:02,814][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:21:03,314][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:21:03,814][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:21:04,313][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:21:04,813][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:21:05,308][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:21:05,818][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:21:06,319][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:21:06,819][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:21:07,319][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:21:07,818][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:21:08,318][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:21:08,817][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:21:09,317][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:21:09,816][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:21:10,316][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:21:10,815][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:21:11,317][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:21:11,816][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:21:12,316][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:21:12,816][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:21:13,314][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:21:13,813][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:21:14,312][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:21:14,811][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:21:15,310][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:21:15,808][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:21:16,307][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:21:16,806][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:21:17,303][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:21:17,802][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:21:18,304][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:21:18,802][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:21:19,300][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:21:19,798][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:21:20,297][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:21:20,796][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:21:21,294][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:21:21,793][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:21:22,293][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:21:22,792][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:21:23,291][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:21:23,790][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:21:24,289][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:21:24,789][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:21:25,287][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:21:25,785][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:21:26,282][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:21:26,779][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:21:27,277][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:21:27,775][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:21:28,273][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:21:28,771][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:21:29,272][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:21:29,771][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:21:30,272][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:21:30,771][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:21:31,270][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:21:31,770][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:21:32,271][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:21:32,772][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:21:33,270][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:21:33,769][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:21:34,267][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:21:34,765][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:21:35,262][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:21:35,759][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:21:36,258][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:21:36,757][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:21:37,255][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:21:37,753][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:21:38,251][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:21:38,748][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:21:39,246][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:21:39,743][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:21:40,241][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:21:40,739][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:21:41,238][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:21:41,736][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:21:42,235][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:21:42,735][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:21:43,234][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:21:43,732][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:21:44,230][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:21:44,728][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:21:45,226][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:21:45,724][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:21:46,223][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:21:46,721][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:21:47,220][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:21:47,718][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:21:48,220][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:21:48,719][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21722 tokens. [2026-03-26 01:21:49,355][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 01:21:50,096][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:21:50,099][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:21:50,101][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:21:50,823][__main__][INFO] - Iteration 223 took 1m 13s (8.74% Gen, 90.28% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 51m 49s. Estimated total time: 61h 38m 57s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 17s, 500 more iterations: 10h 16m 29s. [2026-03-26 01:21:50,825][__main__][INFO] - Starting iteration 223. [2026-03-26 01:21:51,228][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:21:51,229][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:21:56,724][mllm.models.large_language_model_local][WARNING] - Response Proposals: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:21:57,746][__main__][INFO] - Number of regex retries in iteration 223: 1 [2026-03-26 01:21:57,747][__main__][INFO] - agents played in iteration 223 are Alice, Bob [2026-03-26 01:21:58,620][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:21:58,664][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:21:58,665][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:21:58,665][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:21:59,263][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:21:59,717][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:22:00,219][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:22:00,718][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:22:01,216][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:22:01,714][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:22:02,211][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:22:02,708][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:22:03,204][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:22:03,704][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:22:04,203][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:22:04,701][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:22:05,199][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:22:05,697][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:22:06,196][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:22:06,695][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:22:07,194][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:22:07,692][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:22:08,192][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:22:08,689][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:22:09,186][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:22:09,683][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:22:10,181][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:22:10,678][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:22:11,175][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:22:11,672][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:22:12,172][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:22:12,670][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:22:13,168][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:22:13,665][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:22:14,162][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:22:14,659][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:22:15,157][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:22:15,660][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:22:16,161][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:22:16,660][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:22:17,159][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:22:17,658][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:22:18,158][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:22:18,659][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:22:19,157][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:22:19,655][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:22:20,153][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:22:20,651][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:22:21,149][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:22:21,647][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:22:22,145][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:22:22,642][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:22:23,139][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:22:23,637][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:22:24,135][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:22:24,633][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:22:25,133][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:22:25,631][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:22:26,130][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:22:26,628][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:22:27,126][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:22:27,624][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:22:28,124][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:22:28,623][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:22:29,121][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:22:29,619][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:22:30,118][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:22:30,617][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:22:31,115][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:22:31,612][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:22:32,109][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:22:32,606][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:22:33,104][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:22:33,602][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:22:34,100][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:22:34,598][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:22:35,095][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:22:35,594][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:22:36,093][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:22:36,592][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:22:37,091][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:22:37,590][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:22:38,088][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:22:38,587][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:22:39,085][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:22:39,582][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:22:40,078][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:22:40,576][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:22:41,073][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:22:41,573][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:22:42,071][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:22:42,569][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:22:43,067][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:22:43,563][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:22:44,059][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:22:44,557][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:22:45,054][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:22:45,551][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:22:46,049][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:22:46,546][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:22:47,043][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:22:47,542][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:22:48,042][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:22:48,543][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:22:49,042][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:22:49,540][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:22:50,040][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:22:50,539][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:22:51,038][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:22:51,536][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:22:52,034][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:22:52,532][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:22:53,029][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:22:53,527][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:22:54,024][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:22:54,521][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:22:55,020][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:22:55,518][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:22:56,017][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:22:56,515][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:22:57,013][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:22:57,511][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:22:58,009][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:22:58,508][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:22:59,006][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:22:59,504][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:23:00,003][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:23:00,502][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:23:01,001][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:23:01,499][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:23:02,000][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:23:02,499][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:23:02,999][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:23:03,630][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.26%, ΔTime: 00:01:04 [2026-03-26 01:23:04,367][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:23:04,370][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:23:04,371][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:23:05,187][__main__][INFO] - Iteration 224 took 1m 13s (8.81% Gen, 90.08% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 49m 34s. Estimated total time: 61h 37m 57s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 15s, 500 more iterations: 10h 16m 19s. [2026-03-26 01:23:05,189][__main__][INFO] - Starting iteration 224. [2026-03-26 01:23:05,599][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:23:05,600][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:23:07,638][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:23:11,770][__main__][INFO] - Number of regex retries in iteration 224: 1 [2026-03-26 01:23:11,771][__main__][INFO] - agents played in iteration 224 are Alice, Bob [2026-03-26 01:23:12,646][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:23:12,691][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:23:12,691][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:23:12,692][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:23:13,297][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:23:13,751][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:23:14,254][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:23:14,753][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:23:15,252][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:23:15,750][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:23:16,248][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:23:16,747][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:23:17,244][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:23:17,739][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:23:18,236][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:23:18,731][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:23:19,226][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:23:19,721][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:23:20,217][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:23:20,712][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:23:21,207][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:23:21,706][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:23:22,204][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:23:22,701][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:23:23,199][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:23:23,698][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:23:24,195][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:23:24,694][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:23:25,193][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:23:25,694][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:23:26,194][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:23:26,693][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:23:27,192][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:23:27,693][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:23:28,193][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:23:28,693][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:23:29,193][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:23:29,695][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:23:30,195][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:23:30,696][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:23:31,195][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:23:31,695][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:23:32,195][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:23:32,694][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:23:33,194][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:23:33,694][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:23:34,193][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:23:34,692][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:23:35,192][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:23:35,691][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:23:36,190][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:23:36,689][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:23:37,187][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:23:37,685][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:23:38,183][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:23:38,680][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:23:39,178][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:23:39,676][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:23:40,175][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:23:40,673][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:23:41,171][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:23:41,668][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:23:42,166][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:23:42,663][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:23:43,161][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:23:43,660][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:23:44,158][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:23:44,656][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:23:45,155][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:23:45,656][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:23:46,156][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:23:46,653][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:23:47,150][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:23:47,650][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:23:48,148][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:23:48,647][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:23:49,145][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:23:49,640][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:23:50,134][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:23:50,627][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:23:51,121][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:23:51,616][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:23:52,110][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:23:52,604][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:23:53,098][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:23:53,597][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:23:54,095][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:23:54,593][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:23:55,091][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:23:55,589][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:23:56,086][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:23:56,583][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:23:57,080][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:23:57,578][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:23:58,077][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:23:58,576][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:23:59,075][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:23:59,573][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:24:00,071][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:24:00,568][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:24:01,067][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:24:01,565][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:24:02,062][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:24:02,562][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:24:03,061][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:24:03,561][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:24:04,060][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:24:04,559][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:24:05,058][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:24:05,556][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:24:06,055][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:24:06,554][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:24:07,054][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:24:07,552][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:24:08,050][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:24:08,548][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:24:09,046][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:24:09,545][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:24:10,043][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:24:10,541][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:24:11,039][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:24:11,538][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:24:12,036][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:24:12,534][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:24:13,031][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:24:13,529][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:24:14,026][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:24:14,524][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:24:15,022][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:24:15,520][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:24:16,019][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:24:16,517][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:24:17,015][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 01:24:17,653][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:24:18,383][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:24:18,386][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:24:18,387][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:24:19,038][__main__][INFO] - Iteration 225 took 1m 13s (8.40% Gen, 90.71% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 22m 22s. Estimated total time: 61h 11m 59s. Time estimates for 10 more iterations: 12m 14s, 100 more iterations: 2h 2m 23s, 500 more iterations: 10h 11m 59s. [2026-03-26 01:24:19,040][__main__][INFO] - Starting iteration 225. [2026-03-26 01:24:19,441][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:24:19,442][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:24:23,887][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:24:26,090][__main__][INFO] - Number of regex retries in iteration 225: 1 [2026-03-26 01:24:26,091][__main__][INFO] - agents played in iteration 225 are Alice, Bob [2026-03-26 01:24:26,961][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:24:27,006][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:24:27,006][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:24:27,007][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:24:27,610][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:24:28,064][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:24:28,567][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:24:29,065][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:24:29,563][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:24:30,060][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:24:30,558][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:24:31,056][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:24:31,554][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:24:32,055][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:24:32,554][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:24:33,053][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:24:33,551][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:24:34,049][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:24:34,548][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:24:35,047][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:24:35,546][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:24:36,044][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:24:36,542][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:24:37,039][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:24:37,537][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:24:38,035][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:24:38,534][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:24:39,034][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:24:39,533][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:24:40,031][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:24:40,528][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:24:41,026][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:24:41,524][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:24:42,021][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:24:42,520][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:24:43,018][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:24:43,516][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:24:44,015][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:24:44,514][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:24:45,012][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:24:45,509][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:24:46,007][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:24:46,503][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:24:47,001][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:24:47,499][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:24:47,998][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:24:48,498][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:24:48,995][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:24:49,493][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:24:49,991][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:24:50,488][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:24:50,985][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:24:51,482][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:24:51,980][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:24:52,478][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:24:52,975][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:24:53,473][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:24:53,972][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:24:54,469][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:24:54,967][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:24:55,464][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:24:55,963][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:24:56,461][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:24:56,959][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:24:57,458][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:24:57,956][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:24:58,454][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:24:58,952][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:24:59,449][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:24:59,947][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:25:00,444][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:25:00,941][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:25:01,438][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:25:01,936][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:25:02,434][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:25:02,932][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:25:03,431][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:25:03,930][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:25:04,428][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:25:04,926][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:25:05,425][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:25:05,924][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:25:06,423][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:25:06,922][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:25:07,421][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:25:07,920][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:25:08,419][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:25:08,918][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:25:09,418][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:25:09,917][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:25:10,417][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:25:10,915][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:25:11,413][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:25:11,911][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:25:12,409][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:25:12,907][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:25:13,405][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:25:13,903][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:25:14,402][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:25:14,901][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:25:15,401][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:25:15,900][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:25:16,399][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:25:16,897][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:25:17,398][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:25:17,896][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:25:18,395][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:25:18,894][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:25:19,392][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:25:19,890][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:25:20,387][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:25:20,883][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:25:21,380][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:25:21,879][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:25:22,376][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:25:22,873][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:25:23,371][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:25:23,868][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:25:24,366][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:25:24,864][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:25:25,363][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:25:25,862][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:25:26,360][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:25:26,859][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:25:27,357][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:25:27,854][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:25:28,352][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:25:28,850][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:25:29,347][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:25:29,844][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:25:30,342][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:25:30,841][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:25:31,339][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21735 tokens. [2026-03-26 01:25:31,967][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:25:32,713][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:25:32,715][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:25:32,717][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:25:33,528][__main__][INFO] - Iteration 226 took 1m 14s (8.97% Gen, 89.93% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 53m 32s. Estimated total time: 61h 44m 23s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 28s, 500 more iterations: 10h 17m 23s. [2026-03-26 01:25:33,531][__main__][INFO] - Starting iteration 226. [2026-03-26 01:25:33,931][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:25:33,932][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:25:34,518][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:25:35,193][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:25:40,558][__main__][INFO] - Number of regex retries in iteration 226: 2 [2026-03-26 01:25:40,558][__main__][INFO] - agents played in iteration 226 are Alice, Bob [2026-03-26 01:25:41,432][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:25:41,476][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:25:41,477][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:25:41,477][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:25:42,078][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:25:42,536][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:25:43,041][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:25:43,542][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:25:44,043][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:25:44,544][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:25:45,045][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:25:45,546][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:25:46,047][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:25:46,544][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:25:47,041][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:25:47,539][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:25:48,037][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:25:48,535][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:25:49,034][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:25:49,533][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:25:50,031][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:25:50,530][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:25:51,028][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:25:51,526][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:25:52,024][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:25:52,525][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:25:53,023][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:25:53,523][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:25:54,021][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:25:54,522][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:25:55,021][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:25:55,519][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:25:56,019][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:25:56,518][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:25:57,017][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:25:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:25:58,015][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:25:58,514][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:25:59,012][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:25:59,511][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:26:00,009][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:26:00,508][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:26:01,007][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:26:01,506][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:26:02,005][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:26:02,503][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:26:03,001][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:26:03,500][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:26:03,999][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:26:04,496][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:26:04,994][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:26:05,493][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:26:05,991][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:26:06,490][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:26:06,989][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:26:07,488][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:26:07,986][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:26:08,484][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:26:08,984][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:26:09,483][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:26:09,983][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:26:10,481][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:26:10,979][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:26:11,478][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:26:11,977][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:26:12,475][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:26:12,975][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:26:13,475][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:26:13,974][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:26:14,474][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:26:14,977][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:26:15,477][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:26:15,979][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:26:16,480][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:26:16,983][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:26:17,484][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:26:17,985][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:26:18,485][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:26:18,982][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:26:19,479][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:26:19,976][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:26:20,474][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:26:20,971][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:26:21,468][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:26:21,965][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:26:22,463][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:26:22,961][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:26:23,459][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:26:23,958][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:26:24,456][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:26:24,955][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:26:25,454][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:26:25,952][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:26:26,449][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:26:26,948][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:26:27,445][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:26:27,942][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:26:28,441][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:26:28,939][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:26:29,437][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:26:29,935][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:26:30,434][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:26:30,933][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:26:31,431][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:26:31,929][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:26:32,427][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:26:32,925][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:26:33,423][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:26:33,922][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:26:34,419][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:26:34,917][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:26:35,414][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:26:35,911][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:26:36,408][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:26:36,906][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:26:37,403][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:26:37,901][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:26:38,400][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:26:38,900][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:26:39,400][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:26:39,899][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:26:40,399][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:26:40,900][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:26:41,400][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:26:41,900][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:26:42,399][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:26:42,897][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:26:43,396][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:26:43,896][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:26:44,394][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:26:44,893][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:26:45,395][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:26:45,897][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21727 tokens. [2026-03-26 01:26:46,520][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 01:26:47,256][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:26:47,259][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:26:47,260][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:26:47,958][__main__][INFO] - Iteration 227 took 1m 14s (8.95% Gen, 90.10% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 49m 16s. Estimated total time: 61h 41m 22s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 22s, 500 more iterations: 10h 16m 53s. [2026-03-26 01:26:47,961][__main__][INFO] - Starting iteration 227. [2026-03-26 01:26:48,361][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:26:48,362][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:26:54,971][__main__][INFO] - Number of regex retries in iteration 227: 0 [2026-03-26 01:26:54,972][__main__][INFO] - agents played in iteration 227 are Alice, Bob [2026-03-26 01:26:55,846][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:26:55,890][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:26:55,891][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:26:55,892][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:26:56,499][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:26:56,953][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:26:57,453][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:26:57,947][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:26:58,441][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:26:58,935][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:26:59,430][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:26:59,924][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:27:00,419][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:27:01,023][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:27:01,522][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:27:02,020][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:27:02,520][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:27:03,019][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:27:03,518][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:27:04,017][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:27:04,516][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:27:05,019][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:27:05,520][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:27:06,020][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:27:06,520][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:27:07,021][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:27:07,521][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:27:08,020][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:27:08,520][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:27:09,020][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:27:09,520][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:27:10,019][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:27:10,518][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:27:11,017][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:27:11,516][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:27:12,016][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:27:12,515][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:27:13,031][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:27:13,537][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:27:14,038][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:27:14,540][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:27:15,041][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:27:15,543][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:27:16,044][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:27:16,545][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:27:17,044][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:27:17,542][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:27:18,040][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:27:18,541][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:27:19,040][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:27:19,539][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:27:20,038][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:27:20,537][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:27:21,035][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:27:21,534][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:27:22,032][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:27:22,529][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:27:23,027][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:27:23,525][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:27:24,022][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:27:24,521][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:27:25,019][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:27:25,518][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:27:26,016][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:27:26,514][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:27:27,012][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:27:27,509][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:27:28,006][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:27:28,504][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:27:29,000][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:27:29,494][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:27:29,989][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:27:30,484][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:27:30,979][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:27:31,474][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:27:31,968][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:27:32,461][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:27:32,960][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:27:33,459][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:27:33,957][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:27:34,457][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:27:34,956][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:27:35,455][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:27:35,953][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:27:36,451][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:27:36,951][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:27:37,449][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:27:37,949][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:27:38,447][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:27:38,947][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:27:39,446][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:27:39,945][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:27:40,445][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:27:40,943][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:27:41,442][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:27:41,941][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:27:42,440][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:27:42,940][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:27:43,439][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:27:43,938][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:27:44,436][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:27:44,937][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:27:45,438][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:27:45,937][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:27:46,438][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:27:46,939][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:27:47,439][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:27:47,940][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:27:48,442][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:27:48,941][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:27:49,440][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:27:49,939][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:27:50,438][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:27:50,937][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:27:51,435][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:27:51,934][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:27:52,431][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:27:52,929][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:27:53,426][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:27:53,924][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:27:54,422][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:27:54,920][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:27:55,418][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:27:55,915][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:27:56,413][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:27:56,910][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:27:57,407][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:27:57,903][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:27:58,400][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:27:58,898][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:27:59,395][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:27:59,892][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:28:00,388][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 01:28:01,017][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 01:28:01,757][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:28:01,759][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:28:01,761][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:28:02,409][__main__][INFO] - Iteration 228 took 1m 14s (8.93% Gen, 90.20% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 49m 4s. Estimated total time: 61h 42m 24s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 24s, 500 more iterations: 10h 17m 4s. [2026-03-26 01:28:02,411][__main__][INFO] - Starting iteration 228. [2026-03-26 01:28:02,822][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:28:02,822][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:28:09,190][__main__][INFO] - Number of regex retries in iteration 228: 0 [2026-03-26 01:28:09,191][__main__][INFO] - agents played in iteration 228 are Alice, Bob [2026-03-26 01:28:10,075][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:28:10,119][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:28:10,119][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:28:10,120][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:28:10,735][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:28:11,189][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:28:11,691][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:28:12,189][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:28:12,686][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:28:13,184][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:28:13,682][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:28:14,180][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:28:14,678][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:28:15,179][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:28:15,681][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:28:16,182][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:28:16,681][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:28:17,180][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:28:17,678][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:28:18,178][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:28:18,679][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:28:19,179][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:28:19,679][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:28:20,178][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:28:20,678][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:28:21,177][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:28:21,676][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:28:22,176][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:28:22,675][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:28:23,177][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:28:23,678][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:28:24,178][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:28:24,678][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:28:25,179][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:28:25,680][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:28:26,180][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:28:26,681][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:28:27,181][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:28:27,682][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:28:28,182][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:28:28,681][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:28:29,182][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:28:29,682][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:28:30,182][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:28:30,683][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:28:31,182][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:28:31,681][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:28:32,180][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:28:32,679][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:28:33,179][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:28:33,678][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:28:34,177][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:28:34,676][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:28:35,174][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:28:35,672][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:28:36,170][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:28:36,668][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:28:37,166][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:28:37,665][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:28:38,163][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:28:38,662][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:28:39,160][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:28:39,657][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:28:40,153][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:28:40,651][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:28:41,148][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:28:41,644][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:28:42,142][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:28:42,639][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:28:43,137][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:28:43,635][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:28:44,133][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:28:44,629][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:28:45,127][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:28:45,624][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:28:46,121][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:28:46,618][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:28:47,116][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:28:47,615][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:28:48,115][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:28:48,615][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:28:49,113][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:28:49,611][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:28:50,109][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:28:50,607][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:28:51,106][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:28:51,604][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:28:52,103][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:28:52,602][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:28:53,100][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:28:53,598][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:28:54,098][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:28:54,598][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:28:55,099][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:28:55,599][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:28:56,099][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:28:56,598][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:28:57,098][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:28:57,598][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:28:58,098][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:28:58,600][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:28:59,101][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:28:59,602][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:29:00,102][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:29:00,601][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:29:01,101][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:29:01,602][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:29:02,103][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:29:02,604][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:29:03,103][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:29:03,602][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:29:04,101][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:29:04,600][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:29:05,101][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:29:05,601][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:29:06,102][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:29:06,601][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:29:07,101][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:29:07,601][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:29:08,100][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:29:08,599][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:29:09,099][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:29:09,598][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:29:10,097][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:29:10,596][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:29:11,094][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:29:11,591][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:29:12,088][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:29:12,585][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:29:13,082][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:29:13,581][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:29:14,078][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:29:14,575][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:29:15,222][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:29:15,964][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:29:15,966][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:29:15,968][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:29:16,627][__main__][INFO] - Iteration 229 took 1m 13s (8.63% Gen, 90.48% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 35m 44s. Estimated total time: 61h 30m 18s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 0s, 500 more iterations: 10h 15m 3s. [2026-03-26 01:29:16,629][__main__][INFO] - Starting iteration 229. [2026-03-26 01:29:17,033][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:29:17,034][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:29:23,227][__main__][INFO] - Number of regex retries in iteration 229: 0 [2026-03-26 01:29:23,228][__main__][INFO] - agents played in iteration 229 are Alice, Bob [2026-03-26 01:29:24,097][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:29:24,142][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:29:24,142][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:29:24,143][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:29:24,748][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:29:25,205][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:29:25,709][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:29:26,209][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:29:26,709][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:29:27,208][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:29:27,707][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:29:28,207][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:29:28,707][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:29:29,207][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:29:29,706][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:29:30,205][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:29:30,704][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:29:31,203][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:29:31,702][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:29:32,201][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:29:32,700][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:29:33,199][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:29:33,698][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:29:34,197][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:29:34,696][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:29:35,196][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:29:35,696][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:29:36,195][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:29:36,695][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:29:37,193][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:29:37,690][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:29:38,187][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:29:38,683][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:29:39,181][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:29:39,678][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:29:40,176][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:29:40,675][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:29:41,175][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:29:41,674][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:29:42,172][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:29:42,671][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:29:43,169][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:29:43,667][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:29:44,165][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:29:44,663][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:29:45,163][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:29:45,664][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:29:46,164][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:29:46,662][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:29:47,161][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:29:47,659][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:29:48,159][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:29:48,659][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:29:49,161][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:29:49,660][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:29:50,159][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:29:50,659][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:29:51,159][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:29:51,659][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:29:52,159][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:29:52,658][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:29:53,156][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:29:53,654][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:29:54,152][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:29:54,649][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:29:55,148][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:29:55,646][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:29:56,144][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:29:56,642][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:29:57,141][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:29:57,640][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:29:58,139][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:29:58,638][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:29:59,138][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:29:59,637][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:30:00,137][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:30:00,636][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:30:01,135][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:30:01,633][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:30:02,132][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:30:02,630][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:30:03,129][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:30:03,627][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:30:04,125][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:30:04,624][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:30:05,123][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:30:05,621][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:30:06,120][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:30:06,620][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:30:07,120][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:30:07,618][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:30:08,116][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:30:08,616][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:30:09,114][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:30:09,613][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:30:10,110][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:30:10,607][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:30:11,104][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:30:11,601][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:30:12,099][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:30:12,597][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:30:13,096][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:30:13,596][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:30:14,094][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:30:14,592][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:30:15,091][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:30:15,589][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:30:16,087][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:30:16,585][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:30:17,084][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:30:17,582][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:30:18,081][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:30:18,582][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:30:19,081][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:30:19,580][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:30:20,080][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:30:20,578][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:30:21,078][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:30:21,576][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:30:22,075][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:30:22,576][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:30:23,075][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:30:23,577][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:30:24,077][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:30:24,576][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:30:25,075][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:30:25,573][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:30:26,072][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:30:26,570][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:30:27,069][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:30:27,568][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:30:28,066][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:30:28,566][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:30:29,197][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 01:30:29,947][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:30:29,949][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:30:29,951][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:30:30,610][__main__][INFO] - Iteration 230 took 1m 13s (8.42% Gen, 90.68% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 23m 4s. Estimated total time: 61h 18m 52s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 37s, 500 more iterations: 10h 13m 8s. [2026-03-26 01:30:30,612][__main__][INFO] - Starting iteration 230. [2026-03-26 01:30:31,011][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:30:31,012][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:30:37,525][__main__][INFO] - Number of regex retries in iteration 230: 0 [2026-03-26 01:30:37,526][__main__][INFO] - agents played in iteration 230 are Alice, Bob [2026-03-26 01:30:38,397][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:30:38,442][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:30:38,443][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:30:38,443][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:30:39,046][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:30:39,501][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:30:40,004][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:30:40,503][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:30:41,002][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:30:41,502][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:30:42,002][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:30:42,501][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:30:42,999][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:30:43,498][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:30:43,997][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:30:44,496][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:30:44,996][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:30:45,495][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:30:45,994][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:30:46,493][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:30:46,991][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:30:47,490][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:30:47,988][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:30:48,488][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:30:48,987][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:30:49,486][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:30:49,984][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:30:50,483][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:30:50,981][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:30:51,481][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:30:51,979][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:30:52,477][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:30:52,977][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:30:53,477][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:30:53,976][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:30:54,474][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:30:54,973][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:30:55,471][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:30:55,969][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:30:56,467][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:30:56,966][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:30:57,465][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:30:57,964][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:30:58,463][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:30:58,962][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:30:59,460][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:30:59,956][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:31:00,454][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:31:00,951][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:31:01,448][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:31:01,945][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:31:02,443][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:31:02,941][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:31:03,441][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:31:03,940][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:31:04,439][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:31:04,939][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:31:05,438][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:31:05,937][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:31:06,437][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:31:06,936][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:31:07,434][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:31:07,932][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:31:08,429][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:31:08,927][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:31:09,427][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:31:09,925][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:31:10,423][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:31:10,922][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:31:11,422][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:31:11,921][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:31:12,420][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:31:12,920][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:31:13,419][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:31:13,918][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:31:14,416][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:31:14,917][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:31:15,417][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:31:15,918][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:31:16,417][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:31:16,917][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:31:17,416][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:31:17,916][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:31:18,417][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:31:18,916][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:31:19,414][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:31:19,912][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:31:20,410][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:31:20,910][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:31:21,408][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:31:21,908][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:31:22,406][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:31:22,905][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:31:23,404][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:31:23,903][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:31:24,401][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:31:24,900][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:31:25,400][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:31:25,899][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:31:26,397][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:31:26,895][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:31:27,395][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:31:27,894][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:31:28,394][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:31:28,894][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:31:29,394][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:31:29,893][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:31:30,391][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:31:30,889][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:31:31,387][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:31:31,884][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:31:32,382][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:31:32,881][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:31:33,379][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:31:33,877][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:31:34,374][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:31:34,872][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:31:35,371][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:31:35,869][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:31:36,368][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:31:36,867][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:31:37,366][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:31:37,865][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:31:38,364][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:31:38,863][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:31:39,361][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:31:39,859][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:31:40,358][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:31:40,857][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:31:41,356][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:31:41,854][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:31:42,352][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:31:42,850][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 01:31:43,478][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:31:44,354][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:31:44,356][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:31:44,358][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:31:45,056][__main__][INFO] - Iteration 231 took 1m 14s (8.80% Gen, 90.26% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 45m 14s. Estimated total time: 61h 42m 17s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 24s, 500 more iterations: 10h 17m 2s. [2026-03-26 01:31:45,058][__main__][INFO] - Starting iteration 231. [2026-03-26 01:31:45,461][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:31:45,461][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:31:51,684][__main__][INFO] - Number of regex retries in iteration 231: 0 [2026-03-26 01:31:51,685][__main__][INFO] - agents played in iteration 231 are Alice, Bob [2026-03-26 01:31:52,561][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:31:52,605][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:31:52,606][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:31:52,607][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:31:53,206][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:31:53,659][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:31:54,162][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:31:54,661][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:31:55,160][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:31:55,658][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:31:56,156][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:31:56,654][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:31:57,152][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:31:57,652][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:31:58,152][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:31:58,653][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:31:59,152][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:31:59,651][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:32:00,150][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:32:00,650][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:32:01,149][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:32:01,647][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:32:02,144][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:32:02,643][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:32:03,141][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:32:03,639][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:32:04,137][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:32:04,636][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:32:05,134][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:32:05,633][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:32:06,132][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:32:06,631][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:32:07,129][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:32:07,628][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:32:08,127][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:32:08,625][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:32:09,124][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:32:09,622][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:32:10,122][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:32:10,620][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:32:11,120][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:32:11,619][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:32:12,117][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:32:12,615][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:32:13,114][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:32:13,614][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:32:14,113][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:32:14,611][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:32:15,110][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:32:15,608][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:32:16,107][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:32:16,605][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:32:17,103][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:32:17,603][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:32:18,102][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:32:18,604][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:32:19,103][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:32:19,602][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:32:20,101][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:32:20,600][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:32:21,099][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:32:21,599][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:32:22,098][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:32:22,597][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:32:23,097][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:32:23,596][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:32:24,097][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:32:24,597][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:32:25,099][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:32:25,597][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:32:26,095][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:32:26,596][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:32:27,094][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:32:27,592][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:32:28,090][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:32:28,588][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:32:29,086][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:32:29,585][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:32:30,085][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:32:30,584][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:32:31,082][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:32:31,581][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:32:32,080][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:32:32,579][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:32:33,080][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:32:33,578][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:32:34,077][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:32:34,576][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:32:35,073][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:32:35,571][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:32:36,069][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:32:36,566][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:32:37,064][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:32:37,563][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:32:38,062][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:32:38,561][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:32:39,059][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:32:39,557][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:32:40,055][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:32:40,554][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:32:41,053][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:32:41,552][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:32:42,051][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:32:42,550][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:32:43,048][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:32:43,547][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:32:44,045][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:32:44,544][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:32:45,043][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:32:45,544][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:32:46,046][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:32:46,545][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:32:47,044][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:32:47,542][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:32:48,042][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:32:48,543][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:32:49,042][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:32:49,542][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:32:50,040][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:32:50,539][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:32:51,037][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:32:51,536][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:32:52,033][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:32:52,532][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:32:53,031][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:32:53,529][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:32:54,028][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:32:54,527][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:32:55,027][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:32:55,527][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:32:56,025][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:32:56,525][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:32:57,024][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 01:32:57,655][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:32:58,391][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:32:58,393][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:32:58,395][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:32:59,046][__main__][INFO] - Iteration 232 took 1m 13s (8.46% Gen, 90.66% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 21m 1s. Estimated total time: 61h 19m 18s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 38s, 500 more iterations: 10h 13m 13s. [2026-03-26 01:32:59,049][__main__][INFO] - Starting iteration 232. [2026-03-26 01:32:59,457][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:32:59,458][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:33:02,697][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:33:04,292][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:33:05,620][__main__][INFO] - Number of regex retries in iteration 232: 2 [2026-03-26 01:33:05,621][__main__][INFO] - agents played in iteration 232 are Alice, Bob [2026-03-26 01:33:06,508][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:33:06,829][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:33:06,830][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:33:06,830][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:33:07,440][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:33:07,898][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:33:08,402][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:33:08,903][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:33:09,403][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:33:09,903][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:33:10,403][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:33:10,902][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:33:11,402][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:33:11,903][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:33:12,404][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:33:12,905][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:33:13,407][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:33:13,908][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:33:14,407][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:33:14,907][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:33:15,407][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:33:15,907][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:33:16,407][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:33:16,906][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:33:17,407][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:33:17,907][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:33:18,407][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:33:18,908][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:33:19,408][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:33:19,907][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:33:20,406][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:33:20,906][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:33:21,405][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:33:21,904][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:33:22,403][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:33:22,902][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:33:23,400][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:33:23,899][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:33:24,398][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:33:24,897][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:33:25,396][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:33:25,894][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:33:26,391][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:33:26,888][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:33:27,386][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:33:27,884][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:33:28,383][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:33:28,882][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:33:29,381][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:33:29,879][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:33:30,377][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:33:30,877][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:33:31,376][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:33:31,875][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:33:32,376][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:33:32,875][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:33:33,375][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:33:33,873][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:33:34,371][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:33:34,868][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:33:35,366][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:33:35,866][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:33:36,366][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:33:36,866][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:33:37,366][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:33:37,865][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:33:38,365][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:33:38,864][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:33:39,364][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:33:39,863][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:33:40,362][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:33:40,862][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:33:41,360][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:33:41,860][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:33:42,358][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:33:42,857][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:33:43,357][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:33:43,857][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:33:44,357][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:33:44,857][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:33:45,357][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:33:45,856][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:33:46,356][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:33:46,858][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:33:47,358][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:33:47,858][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:33:48,360][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:33:48,861][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:33:49,361][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:33:49,861][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:33:50,361][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:33:50,861][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:33:51,361][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:33:51,860][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:33:52,359][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:33:52,858][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:33:53,356][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:33:53,857][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:33:54,356][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:33:54,856][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:33:55,356][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:33:55,853][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:33:56,351][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:33:56,848][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:33:57,344][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:33:57,841][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:33:58,338][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:33:58,836][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:33:59,333][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:33:59,832][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:34:00,330][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:34:00,828][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:34:01,326][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:34:01,824][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:34:02,323][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:34:02,822][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:34:03,320][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:34:03,819][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:34:04,317][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:34:04,815][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:34:05,313][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:34:05,811][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:34:06,309][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:34:06,808][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:34:07,306][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:34:07,806][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:34:08,304][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:34:08,803][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:34:09,302][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:34:09,801][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:34:10,300][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:34:10,799][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:34:11,299][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:34:11,938][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:34:12,818][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:34:12,821][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:34:12,822][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:34:13,736][__main__][INFO] - Iteration 233 took 1m 14s (8.30% Gen, 90.47% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 54m 26s. Estimated total time: 61h 53m 58s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 59s. [2026-03-26 01:34:13,738][__main__][INFO] - Starting iteration 233. [2026-03-26 01:34:14,137][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:34:14,137][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:34:20,586][__main__][INFO] - Number of regex retries in iteration 233: 0 [2026-03-26 01:34:20,587][__main__][INFO] - agents played in iteration 233 are Alice, Bob [2026-03-26 01:34:21,462][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:34:21,507][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:34:21,507][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:34:21,508][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:34:22,119][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:34:22,573][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:34:23,075][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:34:23,572][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:34:24,070][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:34:24,567][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:34:25,066][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:34:25,564][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:34:26,063][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:34:26,564][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:34:27,063][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:34:27,561][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:34:28,061][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:34:28,559][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:34:29,058][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:34:29,558][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:34:30,057][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:34:30,556][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:34:31,054][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:34:31,551][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:34:32,049][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:34:32,546][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:34:33,044][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:34:33,542][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:34:34,040][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:34:34,538][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:34:35,037][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:34:35,534][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:34:36,031][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:34:36,528][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:34:37,026][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:34:37,524][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:34:38,022][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:34:38,520][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:34:39,018][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:34:39,517][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:34:40,014][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:34:40,512][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:34:41,009][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:34:41,506][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:34:42,004][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:34:42,504][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:34:43,002][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:34:43,499][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:34:43,998][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:34:44,496][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:34:44,993][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:34:45,490][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:34:45,987][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:34:46,485][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:34:46,984][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:34:47,483][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:34:47,983][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:34:48,484][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:34:48,983][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:34:49,482][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:34:49,981][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:34:50,482][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:34:50,981][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:34:51,480][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:34:51,979][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:34:52,478][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:34:52,977][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:34:53,475][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:34:53,974][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:34:54,472][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:34:54,970][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:34:55,468][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:34:55,966][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:34:56,464][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:34:56,962][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:34:57,460][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:34:57,958][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:34:58,458][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:34:58,958][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:34:59,457][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:34:59,956][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:35:00,455][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:35:00,953][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:35:01,450][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:35:01,948][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:35:02,445][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:35:02,943][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:35:03,440][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:35:03,939][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:35:04,436][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:35:04,934][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:35:05,431][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:35:05,929][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:35:06,426][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:35:06,923][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:35:07,421][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:35:07,920][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:35:08,418][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:35:08,916][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:35:09,414][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:35:09,911][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:35:10,409][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:35:10,907][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:35:11,405][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:35:11,903][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:35:12,402][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:35:12,900][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:35:13,398][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:35:13,896][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:35:14,394][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:35:14,892][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:35:15,388][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:35:15,887][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:35:16,385][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:35:16,883][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:35:17,380][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:35:17,878][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:35:18,379][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:35:18,878][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:35:19,376][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:35:19,875][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:35:20,374][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:35:20,872][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:35:21,370][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:35:21,869][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:35:22,367][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:35:22,865][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:35:23,364][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:35:23,862][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:35:24,361][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:35:24,860][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:35:25,359][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:35:25,858][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 01:35:26,496][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 01:35:27,230][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:35:27,232][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:35:27,234][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:35:27,889][__main__][INFO] - Iteration 234 took 1m 13s (8.74% Gen, 90.37% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 26m 52s. Estimated total time: 61h 27m 38s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 55s, 500 more iterations: 10h 14m 36s. [2026-03-26 01:35:27,891][__main__][INFO] - Starting iteration 234. [2026-03-26 01:35:28,292][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:35:28,293][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:35:34,606][__main__][INFO] - Number of regex retries in iteration 234: 0 [2026-03-26 01:35:34,607][__main__][INFO] - agents played in iteration 234 are Alice, Bob [2026-03-26 01:35:35,483][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:35:35,806][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:35:35,807][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:35:35,808][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:35:36,403][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:35:36,856][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:35:37,358][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:35:37,854][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:35:38,351][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:35:38,848][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:35:39,345][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:35:39,842][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:35:40,340][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:35:40,841][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:35:41,340][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:35:41,839][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:35:42,338][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:35:42,838][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:35:43,339][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:35:43,838][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:35:44,338][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:35:44,836][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:35:45,335][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:35:45,836][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:35:46,333][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:35:46,832][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:35:47,330][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:35:47,827][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:35:48,327][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:35:48,826][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:35:49,323][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:35:49,820][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:35:50,317][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:35:50,814][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:35:51,311][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:35:51,808][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:35:52,307][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:35:52,811][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:35:53,313][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:35:53,815][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:35:54,316][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:35:54,819][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:35:55,322][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:35:55,824][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:35:56,325][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:35:56,823][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:35:57,321][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:35:57,819][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:35:58,318][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:35:58,818][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:35:59,316][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:35:59,815][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:36:00,312][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:36:00,810][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:36:01,308][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:36:01,806][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:36:02,303][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:36:02,802][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:36:03,300][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:36:03,798][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:36:04,297][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:36:04,796][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:36:05,294][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:36:05,792][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:36:06,291][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:36:06,790][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:36:07,288][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:36:07,787][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:36:08,286][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:36:08,783][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:36:09,281][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:36:09,779][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:36:10,276][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:36:10,774][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:36:11,271][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:36:11,768][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:36:12,266][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:36:12,765][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:36:13,263][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:36:13,763][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:36:14,261][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:36:14,761][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:36:15,260][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:36:15,761][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:36:16,260][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:36:16,757][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:36:17,254][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:36:17,752][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:36:18,250][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:36:18,748][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:36:19,246][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:36:19,744][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:36:20,242][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:36:20,739][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:36:21,237][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:36:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:36:22,231][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:36:22,729][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:36:23,226][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:36:23,724][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:36:24,222][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:36:24,722][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:36:25,224][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:36:25,726][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:36:26,227][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:36:26,730][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:36:27,232][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:36:27,732][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:36:28,233][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:36:28,731][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:36:29,229][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:36:29,726][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:36:30,224][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:36:30,722][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:36:31,221][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:36:31,719][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:36:32,217][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:36:32,715][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:36:33,213][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:36:33,711][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:36:34,209][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:36:34,707][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:36:35,205][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:36:35,704][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:36:36,203][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:36:36,702][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:36:37,200][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:36:37,700][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:36:38,199][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:36:38,698][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:36:39,198][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:36:39,698][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:36:40,197][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21740 tokens. [2026-03-26 01:36:40,824][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 01:36:41,736][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:36:41,738][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:36:41,740][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:36:42,453][__main__][INFO] - Iteration 235 took 1m 14s (8.51% Gen, 90.52% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 46m 5s. Estimated total time: 61h 48m 6s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 36s, 500 more iterations: 10h 18m 1s. [2026-03-26 01:36:42,456][__main__][INFO] - Starting iteration 235. [2026-03-26 01:36:42,861][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:36:42,862][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:36:43,447][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:36:44,906][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:36:46,675][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:36:49,720][__main__][INFO] - Number of regex retries in iteration 235: 3 [2026-03-26 01:36:49,721][__main__][INFO] - agents played in iteration 235 are Alice, Bob [2026-03-26 01:36:50,595][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:36:50,639][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:36:50,640][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:36:50,640][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:36:51,241][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:36:51,696][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:36:52,201][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:36:52,701][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:36:53,200][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:36:53,699][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:36:54,199][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:36:54,698][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:36:55,197][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:36:55,697][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:36:56,196][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:36:56,694][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:36:57,193][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:36:57,691][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:36:58,190][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:36:58,688][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:36:59,185][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:36:59,683][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:37:00,182][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:37:00,680][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:37:01,179][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:37:01,677][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:37:02,176][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:37:02,674][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:37:03,171][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:37:03,668][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:37:04,165][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:37:04,663][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:37:05,160][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:37:05,659][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:37:06,157][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:37:06,656][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:37:07,153][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:37:07,652][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:37:08,151][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:37:08,648][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:37:09,146][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:37:09,643][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:37:10,142][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:37:10,640][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:37:11,139][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:37:11,637][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:37:12,136][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:37:12,634][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:37:13,131][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:37:13,629][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:37:14,126][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:37:14,624][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:37:15,122][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:37:15,626][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:37:16,126][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:37:16,624][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:37:17,123][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:37:17,621][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:37:18,121][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:37:18,619][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:37:19,118][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:37:19,618][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:37:20,116][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:37:20,615][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:37:21,114][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:37:21,613][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:37:22,113][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:37:22,613][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:37:23,112][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:37:23,611][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:37:24,110][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:37:24,609][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:37:25,107][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:37:25,606][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:37:26,106][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:37:26,605][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:37:27,104][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:37:27,602][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:37:28,101][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:37:28,600][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:37:29,099][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:37:29,598][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:37:30,097][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:37:30,595][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:37:31,093][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:37:31,591][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:37:32,089][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:37:32,586][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:37:33,084][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:37:33,582][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:37:34,080][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:37:34,578][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:37:35,076][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:37:35,573][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:37:36,070][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:37:36,567][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:37:37,064][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:37:37,561][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:37:38,059][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:37:38,556][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:37:39,054][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:37:39,552][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:37:40,049][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:37:40,546][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:37:41,043][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:37:41,541][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:37:42,038][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:37:42,535][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:37:43,032][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:37:43,530][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:37:44,029][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:37:44,527][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:37:45,025][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:37:45,525][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:37:46,023][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:37:46,522][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:37:47,021][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:37:47,521][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:37:48,020][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:37:48,522][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:37:49,022][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:37:49,521][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:37:50,020][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:37:50,519][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:37:51,020][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:37:51,519][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:37:52,019][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:37:52,519][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:37:53,018][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:37:53,516][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:37:54,015][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:37:54,514][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:37:55,013][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 01:37:55,641][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 01:37:56,401][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:37:56,403][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:37:56,405][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:37:57,056][__main__][INFO] - Iteration 236 took 1m 14s (9.24% Gen, 89.88% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 46m 31s. Estimated total time: 61h 49m 46s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 17s. [2026-03-26 01:37:57,058][__main__][INFO] - Starting iteration 236. [2026-03-26 01:37:57,458][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:37:57,458][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:37:59,702][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:38:04,378][__main__][INFO] - Number of regex retries in iteration 236: 1 [2026-03-26 01:38:04,379][__main__][INFO] - agents played in iteration 236 are Alice, Bob [2026-03-26 01:38:05,253][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:38:05,298][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:38:05,298][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:38:05,299][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:38:05,912][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:38:06,366][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:38:06,868][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:38:07,366][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:38:07,864][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:38:08,361][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:38:08,860][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:38:09,358][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:38:09,857][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:38:10,354][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:38:10,852][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:38:11,349][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:38:11,846][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:38:12,343][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:38:12,841][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:38:13,339][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:38:13,838][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:38:14,338][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:38:14,839][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:38:15,338][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:38:15,836][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:38:16,335][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:38:16,833][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:38:17,331][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:38:17,830][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:38:18,330][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:38:18,829][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:38:19,329][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:38:19,828][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:38:20,326][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:38:20,825][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:38:21,323][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:38:21,822][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:38:22,319][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:38:22,817][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:38:23,315][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:38:23,812][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:38:24,309][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:38:24,807][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:38:25,306][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:38:25,803][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:38:26,298][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:38:26,794][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:38:27,289][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:38:27,783][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:38:28,277][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:38:28,771][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:38:29,266][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:38:29,760][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:38:30,257][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:38:30,755][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:38:31,253][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:38:31,750][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:38:32,249][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:38:32,747][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:38:33,246][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:38:33,744][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:38:34,245][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:38:34,745][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:38:35,245][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:38:35,745][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:38:36,245][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:38:36,745][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:38:37,245][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:38:37,744][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:38:38,241][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:38:38,740][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:38:39,238][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:38:39,738][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:38:40,237][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:38:40,736][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:38:41,234][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:38:41,732][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:38:42,229][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:38:42,726][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:38:43,225][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:38:43,722][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:38:44,221][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:38:44,719][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:38:45,219][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:38:45,721][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:38:46,220][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:38:46,719][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:38:47,217][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:38:47,717][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:38:48,217][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:38:48,719][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:38:49,219][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:38:49,717][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:38:50,215][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:38:50,713][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:38:51,211][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:38:51,710][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:38:52,208][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:38:52,708][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:38:53,206][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:38:53,705][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:38:54,203][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:38:54,701][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:38:55,200][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:38:55,698][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:38:56,196][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:38:56,693][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:38:57,190][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:38:57,687][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:38:58,183][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:38:58,679][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:38:59,176][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:38:59,670][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:39:00,164][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:39:00,658][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:39:01,151][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:39:01,645][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:39:02,142][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:39:02,641][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:39:03,138][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:39:03,636][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:39:04,133][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:39:04,631][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:39:05,129][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:39:05,626][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:39:06,125][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:39:06,625][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:39:07,125][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:39:07,625][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:39:08,124][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:39:08,624][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:39:09,123][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:39:09,623][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 01:39:10,267][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:39:11,007][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:39:11,010][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:39:11,011][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:39:11,669][__main__][INFO] - Iteration 237 took 1m 14s (9.32% Gen, 89.79% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 46m 6s. Estimated total time: 61h 50m 36s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 26s. [2026-03-26 01:39:11,671][__main__][INFO] - Starting iteration 237. [2026-03-26 01:39:12,071][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:39:12,071][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:39:12,663][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:39:18,482][__main__][INFO] - Number of regex retries in iteration 237: 1 [2026-03-26 01:39:18,483][__main__][INFO] - agents played in iteration 237 are Alice, Bob [2026-03-26 01:39:19,362][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:39:19,406][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:39:19,407][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:39:19,407][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:39:20,012][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:39:20,463][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:39:20,966][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:39:21,462][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:39:21,958][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:39:22,453][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:39:22,949][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:39:23,445][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:39:23,940][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:39:24,440][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:39:24,937][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:39:25,435][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:39:25,932][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:39:26,429][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:39:26,926][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:39:27,424][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:39:27,921][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:39:28,420][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:39:28,918][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:39:29,415][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:39:29,913][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:39:30,409][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:39:30,908][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:39:31,407][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:39:31,905][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:39:32,405][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:39:32,905][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:39:33,405][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:39:33,904][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:39:34,404][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:39:34,902][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:39:35,402][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:39:35,901][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:39:36,399][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:39:36,896][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:39:37,394][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:39:37,891][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:39:38,388][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:39:38,886][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:39:39,384][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:39:39,882][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:39:40,380][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:39:40,878][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:39:41,375][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:39:41,872][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:39:42,369][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:39:42,866][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:39:43,364][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:39:43,861][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:39:44,359][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:39:44,857][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:39:45,355][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:39:45,852][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:39:46,350][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:39:46,848][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:39:47,346][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:39:47,844][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:39:48,346][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:39:48,846][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:39:49,345][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:39:49,844][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:39:50,342][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:39:50,841][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:39:51,339][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:39:51,840][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:39:52,336][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:39:52,832][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:39:53,328][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:39:53,825][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:39:54,323][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:39:54,819][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:39:55,316][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:39:55,812][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:39:56,310][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:39:56,808][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:39:57,307][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:39:57,806][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:39:58,304][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:39:58,802][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:39:59,301][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:39:59,800][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:40:00,299][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:40:00,799][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:40:01,298][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:40:01,797][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:40:02,296][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:40:02,794][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:40:03,292][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:40:03,790][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:40:04,288][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:40:04,787][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:40:05,286][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:40:05,786][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:40:06,286][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:40:06,785][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:40:07,285][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:40:07,785][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:40:08,284][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:40:08,782][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:40:09,280][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:40:09,779][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:40:10,278][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:40:10,776][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:40:11,274][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:40:11,772][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:40:12,270][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:40:12,768][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:40:13,266][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:40:13,766][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:40:14,264][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:40:14,762][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:40:15,261][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:40:15,762][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:40:16,263][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:40:16,762][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:40:17,260][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:40:17,760][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:40:18,260][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:40:18,760][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:40:19,259][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:40:19,757][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:40:20,256][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:40:20,756][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:40:21,254][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:40:21,754][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:40:22,253][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:40:22,752][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:40:23,250][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:40:23,748][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:40:24,382][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 01:40:25,183][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:40:25,185][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:40:25,187][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:40:25,976][__main__][INFO] - Iteration 238 took 1m 13s (8.67% Gen, 90.26% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 29m 34s. Estimated total time: 61h 35m 17s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 10s, 500 more iterations: 10h 15m 52s. [2026-03-26 01:40:25,979][__main__][INFO] - Starting iteration 238. [2026-03-26 01:40:26,384][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:40:26,385][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:40:26,981][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:40:28,118][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:40:32,737][__main__][INFO] - Number of regex retries in iteration 238: 2 [2026-03-26 01:40:32,738][__main__][INFO] - agents played in iteration 238 are Alice, Bob [2026-03-26 01:40:33,621][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:40:33,665][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:40:33,666][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:40:33,667][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:40:34,277][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:40:34,732][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:40:35,235][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:40:35,734][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:40:36,234][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:40:36,733][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:40:37,232][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:40:37,732][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:40:38,230][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:40:38,729][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:40:39,227][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:40:39,726][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:40:40,225][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:40:40,722][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:40:41,220][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:40:41,718][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:40:42,216][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:40:42,714][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:40:43,211][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:40:43,709][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:40:44,207][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:40:44,705][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:40:45,203][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:40:45,700][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:40:46,198][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:40:46,695][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:40:47,192][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:40:47,690][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:40:48,189][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:40:48,689][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:40:49,187][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:40:49,685][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:40:50,183][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:40:50,683][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:40:51,182][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:40:51,680][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:40:52,182][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:40:52,681][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:40:53,180][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:40:53,679][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:40:54,177][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:40:54,675][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:40:55,173][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:40:55,670][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:40:56,169][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:40:56,668][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:40:57,166][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:40:57,665][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:40:58,163][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:40:58,661][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:40:59,158][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:40:59,655][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:41:00,152][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:41:00,650][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:41:01,148][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:41:01,645][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:41:02,143][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:41:02,642][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:41:03,140][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:41:03,638][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:41:04,136][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:41:04,634][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:41:05,133][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:41:05,632][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:41:06,130][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:41:06,630][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:41:07,128][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:41:07,628][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:41:08,127][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:41:08,626][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:41:09,125][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:41:09,625][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:41:10,124][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:41:10,622][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:41:11,119][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:41:11,617][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:41:12,114][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:41:12,612][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:41:13,110][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:41:13,608][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:41:14,107][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:41:14,605][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:41:15,103][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:41:15,601][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:41:16,101][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:41:16,600][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:41:17,099][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:41:17,600][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:41:18,101][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:41:18,600][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:41:19,097][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:41:19,594][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:41:20,091][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:41:20,588][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:41:21,086][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:41:21,587][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:41:22,086][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:41:22,584][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:41:23,083][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:41:23,580][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:41:24,078][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:41:24,577][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:41:25,076][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:41:25,575][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:41:26,073][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:41:26,571][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:41:27,069][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:41:27,567][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:41:28,066][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:41:28,563][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:41:29,062][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:41:29,561][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:41:30,060][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:41:30,558][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:41:31,055][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:41:31,551][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:41:32,048][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:41:32,544][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:41:33,042][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:41:33,539][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:41:34,037][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:41:34,537][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:41:35,035][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:41:35,534][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:41:36,032][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:41:36,531][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:41:37,029][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:41:37,528][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:41:38,027][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 01:41:38,664][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 01:41:39,402][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:41:39,405][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:41:39,406][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:41:40,056][__main__][INFO] - Iteration 239 took 1m 13s (8.62% Gen, 90.49% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 16m 38s. Estimated total time: 61h 23m 36s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 47s, 500 more iterations: 10h 13m 56s. [2026-03-26 01:41:40,058][__main__][INFO] - Starting iteration 239. [2026-03-26 01:41:40,460][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:41:40,461][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:41:41,047][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:41:47,504][__main__][INFO] - Number of regex retries in iteration 239: 1 [2026-03-26 01:41:47,505][__main__][INFO] - agents played in iteration 239 are Alice, Bob [2026-03-26 01:41:48,382][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:41:48,427][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:41:48,427][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:41:48,428][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:41:49,034][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:41:49,488][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:41:49,992][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:41:50,490][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:41:50,988][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:41:51,488][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:41:51,987][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:41:52,486][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:41:52,985][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:41:53,484][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:41:53,981][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:41:54,481][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:41:54,980][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:41:55,478][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:41:55,975][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:41:56,472][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:41:56,969][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:41:57,467][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:41:57,965][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:41:58,463][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:41:58,961][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:41:59,460][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:41:59,957][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:42:00,455][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:42:00,953][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:42:01,451][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:42:01,948][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:42:02,446][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:42:02,945][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:42:03,443][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:42:03,941][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:42:04,440][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:42:04,939][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:42:05,437][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:42:05,934][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:42:06,433][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:42:06,931][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:42:07,429][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:42:07,927][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:42:08,425][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:42:08,924][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:42:09,424][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:42:09,923][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:42:10,423][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:42:10,922][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:42:11,422][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:42:11,922][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:42:12,421][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:42:12,921][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:42:13,419][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:42:13,916][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:42:14,414][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:42:14,911][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:42:15,409][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:42:15,907][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:42:16,406][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:42:16,904][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:42:17,402][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:42:17,900][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:42:18,400][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:42:18,899][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:42:19,397][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:42:19,895][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:42:20,392][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:42:20,890][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:42:21,388][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:42:21,886][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:42:22,385][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:42:22,885][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:42:23,383][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:42:23,883][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:42:24,382][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:42:24,881][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:42:25,378][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:42:25,875][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:42:26,372][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:42:26,869][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:42:27,366][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:42:27,863][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:42:28,361][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:42:28,860][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:42:29,359][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:42:29,857][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:42:30,355][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:42:30,851][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:42:31,348][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:42:31,846][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:42:32,344][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:42:32,842][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:42:33,340][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:42:33,838][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:42:34,336][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:42:34,833][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:42:35,331][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:42:35,828][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:42:36,326][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:42:36,825][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:42:37,323][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:42:37,822][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:42:38,321][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:42:38,819][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:42:39,317][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:42:39,815][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:42:40,313][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:42:40,811][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:42:41,310][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:42:41,810][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:42:42,310][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:42:42,809][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:42:43,309][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:42:43,806][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:42:44,305][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:42:44,804][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:42:45,303][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:42:45,801][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:42:46,299][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:42:46,796][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:42:47,293][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:42:47,791][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:42:48,290][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:42:48,788][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:42:49,286][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:42:49,784][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:42:50,282][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:42:50,780][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:42:51,279][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:42:51,776][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:42:52,274][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:42:52,771][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 01:42:53,412][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 01:42:54,149][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:42:54,151][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:42:54,153][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:42:54,788][__main__][INFO] - Iteration 240 took 1m 14s (9.48% Gen, 89.67% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 56h 48m 11s. Estimated total time: 61h 56m 23s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 52s, 500 more iterations: 10h 19m 23s. [2026-03-26 01:42:54,790][__main__][INFO] - Starting iteration 240. [2026-03-26 01:42:55,191][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:42:55,191][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:43:01,562][__main__][INFO] - Number of regex retries in iteration 240: 0 [2026-03-26 01:43:01,563][__main__][INFO] - agents played in iteration 240 are Alice, Bob [2026-03-26 01:43:02,439][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:43:02,483][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:43:02,484][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:43:02,484][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:43:03,095][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:43:03,550][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:43:04,052][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:43:04,550][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:43:05,049][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:43:05,547][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:43:06,046][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:43:06,545][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:43:07,042][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:43:07,542][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:43:08,041][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:43:08,540][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:43:09,039][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:43:09,538][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:43:10,036][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:43:10,535][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:43:11,033][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:43:11,534][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:43:12,033][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:43:12,533][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:43:13,033][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:43:13,532][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:43:14,032][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:43:14,532][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:43:15,032][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:43:15,535][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:43:16,037][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:43:16,537][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:43:17,037][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:43:17,536][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:43:18,037][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:43:18,538][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:43:19,039][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:43:19,538][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:43:20,037][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:43:20,536][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:43:21,035][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:43:21,533][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:43:22,031][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:43:22,530][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:43:23,028][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:43:23,527][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:43:24,028][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:43:24,527][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:43:25,027][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:43:25,527][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:43:26,027][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:43:26,526][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:43:27,026][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:43:27,525][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:43:28,024][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:43:28,524][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:43:29,022][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:43:29,522][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:43:30,021][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:43:30,521][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:43:31,021][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:43:31,522][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:43:32,021][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:43:32,521][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:43:33,020][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:43:33,519][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:43:34,019][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:43:34,520][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:43:35,019][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:43:35,519][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:43:36,017][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:43:36,516][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:43:37,014][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:43:37,512][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:43:38,011][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:43:38,510][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:43:39,009][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:43:39,508][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:43:40,008][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:43:40,507][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:43:41,006][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:43:41,506][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:43:42,006][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:43:42,506][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:43:43,005][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:43:43,505][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:43:44,005][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:43:44,506][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:43:45,007][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:43:45,508][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:43:46,009][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:43:46,509][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:43:47,009][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:43:47,508][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:43:48,008][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:43:48,510][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:43:49,011][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:43:49,511][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:43:50,010][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:43:50,510][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:43:51,009][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:43:51,508][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:43:52,006][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:43:52,506][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:43:53,006][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:43:53,505][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:43:54,003][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:43:54,502][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:43:55,002][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:43:55,503][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:43:56,003][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:43:56,502][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:43:57,002][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:43:57,502][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:43:58,001][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:43:58,502][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:43:59,002][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:43:59,502][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:44:00,002][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:44:00,502][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:44:01,001][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:44:01,499][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:44:01,998][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:44:02,497][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:44:02,996][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:44:03,496][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:44:03,995][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:44:04,495][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:44:04,994][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:44:05,493][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:44:05,992][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:44:06,491][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:44:06,990][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 01:44:07,636][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 01:44:08,378][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:44:08,381][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:44:08,382][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:44:09,085][__main__][INFO] - Iteration 241 took 1m 13s (8.62% Gen, 90.42% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 25m 17s. Estimated total time: 61h 34m 44s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 9s, 500 more iterations: 10h 15m 47s. [2026-03-26 01:44:09,087][__main__][INFO] - Starting iteration 241. [2026-03-26 01:44:09,490][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:44:09,490][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:44:16,055][__main__][INFO] - Number of regex retries in iteration 241: 0 [2026-03-26 01:44:16,056][__main__][INFO] - agents played in iteration 241 are Alice, Bob [2026-03-26 01:44:16,935][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:44:16,980][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:44:16,980][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:44:16,981][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:44:17,579][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:44:18,038][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:44:18,544][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:44:19,044][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:44:19,544][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:44:20,044][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:44:20,544][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:44:21,044][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:44:21,543][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:44:22,043][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:44:22,542][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:44:23,041][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:44:23,539][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:44:24,037][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:44:24,535][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:44:25,033][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:44:25,532][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:44:26,032][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:44:26,530][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:44:27,030][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:44:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:44:28,028][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:44:28,526][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:44:29,025][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:44:29,524][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:44:30,025][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:44:30,524][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:44:31,023][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:44:31,522][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:44:32,021][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:44:32,519][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:44:33,018][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:44:33,516][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:44:34,014][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:44:34,512][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:44:35,010][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:44:35,509][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:44:36,007][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:44:36,505][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:44:37,005][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:44:37,503][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:44:38,001][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:44:38,501][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:44:38,999][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:44:39,496][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:44:39,993][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:44:40,491][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:44:40,988][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:44:41,486][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:44:41,984][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:44:42,484][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:44:42,984][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:44:43,483][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:44:43,981][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:44:44,479][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:44:44,978][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:44:45,479][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:44:45,980][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:44:46,480][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:44:46,978][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:44:47,476][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:44:47,974][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:44:48,474][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:44:48,972][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:44:49,469][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:44:49,968][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:44:50,468][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:44:50,967][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:44:51,467][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:44:51,966][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:44:52,466][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:44:52,965][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:44:53,465][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:44:53,963][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:44:54,461][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:44:54,959][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:44:55,457][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:44:55,953][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:44:56,452][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:44:56,948][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:44:57,446][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:44:57,944][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:44:58,443][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:44:58,942][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:44:59,442][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:44:59,940][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:45:00,438][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:45:00,936][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:45:01,434][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:45:01,932][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:45:02,431][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:45:02,930][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:45:03,428][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:45:03,927][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:45:04,426][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:45:04,924][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:45:05,422][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:45:05,919][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:45:06,418][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:45:06,917][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:45:07,415][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:45:07,914][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:45:08,412][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:45:08,911][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:45:09,409][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:45:09,906][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:45:10,404][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:45:10,903][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:45:11,402][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:45:11,900][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:45:12,398][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:45:12,896][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:45:13,393][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:45:13,891][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:45:14,389][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:45:14,888][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:45:15,388][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:45:15,888][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:45:16,387][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:45:16,886][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:45:17,385][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:45:17,884][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:45:18,384][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:45:18,884][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:45:19,383][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:45:19,881][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:45:20,379][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:45:20,878][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:45:21,376][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21747 tokens. [2026-03-26 01:45:22,009][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:45:22,744][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:45:22,747][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:45:22,748][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:45:23,398][__main__][INFO] - Iteration 242 took 1m 13s (8.88% Gen, 90.24% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 24m 44s. Estimated total time: 61h 35m 25s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 10s, 500 more iterations: 10h 15m 54s. [2026-03-26 01:45:23,400][__main__][INFO] - Starting iteration 242. [2026-03-26 01:45:23,797][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:45:23,798][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:45:24,377][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:45:26,504][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:45:30,284][__main__][INFO] - Number of regex retries in iteration 242: 2 [2026-03-26 01:45:30,285][__main__][INFO] - agents played in iteration 242 are Alice, Bob [2026-03-26 01:45:31,156][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:45:31,200][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:45:31,201][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:45:31,202][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:45:31,807][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:45:32,261][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:45:32,764][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:45:33,262][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:45:33,760][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:45:34,258][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:45:34,756][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:45:35,253][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:45:35,751][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:45:36,250][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:45:36,748][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:45:37,247][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:45:37,744][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:45:38,242][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:45:38,740][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:45:39,237][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:45:39,735][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:45:40,235][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:45:40,735][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:45:41,234][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:45:41,733][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:45:42,232][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:45:42,731][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:45:43,230][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:45:43,730][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:45:44,227][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:45:44,725][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:45:45,224][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:45:45,721][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:45:46,218][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:45:46,716][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:45:47,213][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:45:47,710][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:45:48,209][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:45:48,708][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:45:49,206][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:45:49,705][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:45:50,205][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:45:50,704][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:45:51,202][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:45:51,702][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:45:52,202][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:45:52,703][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:45:53,202][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:45:53,703][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:45:54,202][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:45:54,702][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:45:55,204][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:45:55,703][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:45:56,202][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:45:56,701][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:45:57,201][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:45:57,699][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:45:58,197][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:45:58,695][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:45:59,193][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:45:59,691][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:46:00,191][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:46:00,690][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:46:01,188][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:46:01,687][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:46:02,185][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:46:02,685][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:46:03,183][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:46:03,683][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:46:04,182][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:46:04,681][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:46:05,178][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:46:05,675][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:46:06,173][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:46:06,671][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:46:07,168][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:46:07,666][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:46:08,164][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:46:08,661][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:46:09,160][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:46:09,659][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:46:10,157][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:46:10,655][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:46:11,152][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:46:11,650][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:46:12,148][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:46:12,648][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:46:13,148][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:46:13,648][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:46:14,147][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:46:14,645][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:46:15,145][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:46:15,648][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:46:16,148][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:46:16,647][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:46:17,145][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:46:17,643][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:46:18,142][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:46:18,642][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:46:19,139][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:46:19,637][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:46:20,136][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:46:20,633][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:46:21,131][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:46:21,629][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:46:22,127][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:46:22,624][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:46:23,122][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:46:23,619][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:46:24,119][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:46:24,618][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:46:25,117][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:46:25,617][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:46:26,116][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:46:26,614][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:46:27,113][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:46:27,612][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:46:28,111][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:46:28,610][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:46:29,110][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:46:29,609][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:46:30,108][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:46:30,607][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:46:31,106][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:46:31,605][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:46:32,104][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:46:32,603][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:46:33,102][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:46:33,601][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:46:34,099][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:46:34,597][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:46:35,095][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:46:35,593][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21738 tokens. [2026-03-26 01:46:36,223][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 01:46:36,961][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:46:36,963][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:46:36,965][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:46:37,808][__main__][INFO] - Iteration 243 took 1m 14s (8.76% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 28m 37s. Estimated total time: 61h 40m 33s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 21s, 500 more iterations: 10h 16m 45s. [2026-03-26 01:46:37,810][__main__][INFO] - Starting iteration 243. [2026-03-26 01:46:38,215][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:46:38,216][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:46:42,118][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:46:44,767][__main__][INFO] - Number of regex retries in iteration 243: 1 [2026-03-26 01:46:44,768][__main__][INFO] - agents played in iteration 243 are Alice, Bob [2026-03-26 01:46:45,643][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:46:45,687][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:46:45,688][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:46:45,688][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:46:46,298][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:46:46,754][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:46:47,256][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:46:47,754][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:46:48,254][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:46:48,753][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:46:49,251][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:46:49,749][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:46:50,248][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:46:50,749][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:46:51,247][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:46:51,744][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:46:52,242][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:46:52,739][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:46:53,236][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:46:53,733][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:46:54,230][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:46:54,728][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:46:55,225][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:46:55,722][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:46:56,218][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:46:56,716][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:46:57,213][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:46:57,710][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:46:58,208][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:46:58,706][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:46:59,204][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:46:59,702][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:47:00,201][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:47:00,698][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:47:01,195][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:47:01,692][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:47:02,190][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:47:02,690][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:47:03,188][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:47:03,687][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:47:04,187][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:47:04,686][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:47:05,185][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:47:05,685][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:47:06,184][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:47:06,702][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:47:07,206][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:47:07,706][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:47:08,205][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:47:08,705][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:47:09,206][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:47:09,705][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:47:10,205][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:47:10,703][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:47:11,200][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:47:11,697][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:47:12,193][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:47:12,691][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:47:13,189][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:47:13,687][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:47:14,185][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:47:14,684][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:47:15,183][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:47:15,682][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:47:16,182][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:47:16,682][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:47:17,185][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:47:17,684][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:47:18,186][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:47:18,685][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:47:19,182][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:47:19,679][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:47:20,176][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:47:20,672][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:47:21,170][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:47:21,668][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:47:22,165][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:47:22,663][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:47:23,162][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:47:23,661][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:47:24,159][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:47:24,658][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:47:25,156][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:47:25,655][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:47:26,154][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:47:26,652][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:47:27,150][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:47:27,648][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:47:28,145][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:47:28,643][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:47:29,142][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:47:29,642][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:47:30,141][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:47:30,640][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:47:31,137][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:47:31,635][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:47:32,133][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:47:32,630][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:47:33,128][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:47:33,626][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:47:34,123][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:47:34,623][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:47:35,122][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:47:35,621][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:47:36,120][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:47:36,619][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:47:37,117][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:47:37,616][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:47:38,116][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:47:38,616][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:47:39,115][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:47:39,614][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:47:40,112][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:47:40,611][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:47:41,111][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:47:41,611][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:47:42,111][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:47:42,610][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:47:43,108][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:47:43,605][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:47:44,102][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:47:44,599][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:47:45,096][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:47:45,596][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:47:46,094][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:47:46,594][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:47:47,094][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:47:47,593][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:47:48,093][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:47:48,593][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:47:49,092][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:47:49,591][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:47:50,091][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 01:47:50,733][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 01:47:51,471][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:47:51,473][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:47:51,475][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:47:52,304][__main__][INFO] - Iteration 244 took 1m 14s (8.84% Gen, 90.04% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 31m 17s. Estimated total time: 61h 44m 27s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 28s, 500 more iterations: 10h 17m 24s. [2026-03-26 01:47:52,306][__main__][INFO] - Starting iteration 244. [2026-03-26 01:47:52,704][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:47:52,705][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:47:59,169][__main__][INFO] - Number of regex retries in iteration 244: 0 [2026-03-26 01:47:59,170][__main__][INFO] - agents played in iteration 244 are Alice, Bob [2026-03-26 01:48:00,046][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:48:00,090][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:48:00,091][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:48:00,092][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:48:00,699][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:48:01,156][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:48:01,660][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:48:02,160][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:48:02,659][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:48:03,158][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:48:03,657][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:48:04,156][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:48:04,656][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:48:05,154][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:48:05,651][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:48:06,150][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:48:06,647][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:48:07,145][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:48:07,643][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:48:08,142][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:48:08,641][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:48:09,141][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:48:09,639][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:48:10,138][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:48:10,635][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:48:11,133][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:48:11,630][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:48:12,128][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:48:12,627][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:48:13,124][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:48:13,622][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:48:14,119][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:48:14,615][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:48:15,111][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:48:15,608][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:48:16,105][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:48:16,604][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:48:17,102][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:48:17,602][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:48:18,101][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:48:18,601][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:48:19,098][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:48:19,595][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:48:20,091][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:48:20,588][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:48:21,085][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:48:21,584][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:48:22,083][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:48:22,580][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:48:23,076][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:48:23,572][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:48:24,069][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:48:24,566][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:48:25,062][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:48:25,559][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:48:26,057][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:48:26,554][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:48:27,052][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:48:27,549][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:48:28,048][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:48:28,546][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:48:29,044][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:48:29,544][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:48:30,042][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:48:30,541][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:48:31,041][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:48:31,539][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:48:32,037][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:48:32,534][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:48:33,033][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:48:33,532][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:48:34,030][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:48:34,529][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:48:35,028][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:48:35,527][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:48:36,026][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:48:36,525][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:48:37,022][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:48:37,519][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:48:38,017][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:48:38,514][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:48:39,012][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:48:39,511][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:48:40,009][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:48:40,507][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:48:41,005][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:48:41,502][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:48:42,000][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:48:42,497][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:48:42,994][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:48:43,492][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:48:43,989][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:48:44,487][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:48:44,984][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:48:45,482][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:48:45,978][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:48:46,476][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:48:46,973][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:48:47,470][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:48:47,968][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:48:48,468][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:48:48,965][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:48:49,463][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:48:49,959][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:48:50,456][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:48:50,954][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:48:51,450][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:48:51,947][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:48:52,445][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:48:52,944][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:48:53,442][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:48:53,940][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:48:54,438][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:48:54,935][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:48:55,434][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:48:55,931][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:48:56,429][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:48:56,928][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:48:57,426][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:48:57,924][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:48:58,423][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:48:58,921][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:48:59,418][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:48:59,915][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:49:00,413][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:49:00,911][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:49:01,409][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:49:01,909][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:49:02,407][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:49:02,905][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:49:03,403][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:49:03,901][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:49:04,399][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 01:49:05,030][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 01:49:05,765][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:49:05,767][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:49:05,769][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:49:06,545][__main__][INFO] - Iteration 245 took 1m 13s (8.76% Gen, 90.19% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 17m 40s. Estimated total time: 61h 32m 4s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 20s. [2026-03-26 01:49:06,550][__main__][INFO] - Starting iteration 245. [2026-03-26 01:49:06,949][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:49:06,950][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:49:12,847][__main__][INFO] - Number of regex retries in iteration 245: 0 [2026-03-26 01:49:12,849][__main__][INFO] - agents played in iteration 245 are Alice, Bob [2026-03-26 01:49:13,719][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:49:14,042][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:49:14,043][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:49:14,043][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:49:14,641][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:49:15,094][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:49:15,598][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:49:16,098][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:49:16,596][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:49:17,094][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:49:17,591][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:49:18,091][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:49:18,590][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:49:19,090][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:49:19,587][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:49:20,086][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:49:20,584][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:49:21,084][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:49:21,583][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:49:22,082][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:49:22,580][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:49:23,079][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:49:23,578][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:49:24,076][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:49:24,573][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:49:25,073][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:49:25,571][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:49:26,071][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:49:26,569][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:49:27,067][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:49:27,566][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:49:28,064][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:49:28,562][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:49:29,061][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:49:29,559][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:49:30,057][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:49:30,557][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:49:31,055][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:49:31,554][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:49:32,053][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:49:32,552][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:49:33,051][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:49:33,549][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:49:34,048][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:49:34,547][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:49:35,045][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:49:35,543][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:49:36,041][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:49:36,540][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:49:37,038][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:49:37,536][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:49:38,034][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:49:38,532][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:49:39,030][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:49:39,527][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:49:40,023][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:49:40,520][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:49:41,016][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:49:41,513][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:49:42,010][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:49:42,508][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:49:43,006][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:49:43,504][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:49:44,004][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:49:44,502][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:49:45,000][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:49:45,497][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:49:45,996][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:49:46,494][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:49:46,992][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:49:47,488][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:49:47,987][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:49:48,486][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:49:48,983][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:49:49,479][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:49:49,976][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:49:50,472][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:49:50,970][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:49:51,467][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:49:51,965][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:49:52,463][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:49:52,962][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:49:53,459][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:49:53,957][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:49:54,454][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:49:54,952][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:49:55,451][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:49:55,949][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:49:56,448][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:49:56,947][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:49:57,445][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:49:57,943][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:49:58,441][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:49:58,938][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:49:59,436][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:49:59,933][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:50:00,431][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:50:00,930][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:50:01,429][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:50:01,928][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:50:02,426][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:50:02,924][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:50:03,421][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:50:03,921][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:50:04,420][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:50:04,918][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:50:05,417][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:50:05,916][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:50:06,414][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:50:06,912][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:50:07,410][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:50:07,909][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:50:08,409][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:50:08,907][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:50:09,406][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:50:09,905][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:50:10,404][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:50:10,901][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:50:11,397][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:50:11,894][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:50:12,391][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:50:12,889][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:50:13,386][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:50:13,883][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:50:14,380][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:50:14,877][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:50:15,375][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:50:15,872][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:50:16,370][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:50:16,869][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:50:17,367][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:50:17,866][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:50:18,366][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:50:18,992][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-26 01:50:19,728][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:50:19,730][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:50:19,732][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:50:20,521][__main__][INFO] - Iteration 246 took 1m 13s (8.02% Gen, 90.91% Train). Generation: 5s, Training: 1m 6s. Estimated remaining time: 56h 2m 59s. Estimated total time: 61h 18m 37s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 37s, 500 more iterations: 10h 13m 6s. [2026-03-26 01:50:20,523][__main__][INFO] - Starting iteration 246. [2026-03-26 01:50:20,923][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:50:20,924][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:50:27,046][__main__][INFO] - Number of regex retries in iteration 246: 0 [2026-03-26 01:50:27,048][__main__][INFO] - agents played in iteration 246 are Alice, Bob [2026-03-26 01:50:27,924][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:50:28,244][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:50:28,245][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:50:28,246][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:50:28,841][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:50:29,296][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:50:29,799][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:50:30,297][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:50:30,795][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:50:31,295][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:50:31,794][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:50:32,293][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:50:32,792][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:50:33,293][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:50:33,792][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:50:34,290][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:50:34,790][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:50:35,290][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:50:35,790][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:50:36,289][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:50:36,788][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:50:37,287][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:50:37,787][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:50:38,285][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:50:38,786][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:50:39,284][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:50:39,784][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:50:40,283][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:50:40,781][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:50:41,279][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:50:41,778][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:50:42,275][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:50:42,774][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:50:43,272][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:50:43,771][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:50:44,270][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:50:44,768][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:50:45,269][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:50:45,771][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:50:46,272][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:50:46,770][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:50:47,269][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:50:47,768][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:50:48,269][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:50:48,768][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:50:49,267][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:50:49,766][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:50:50,265][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:50:50,765][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:50:51,264][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:50:51,763][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:50:52,261][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:50:52,760][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:50:53,258][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:50:53,755][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:50:54,255][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:50:54,752][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:50:55,252][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:50:55,750][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:50:56,249][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:50:56,747][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:50:57,244][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:50:57,740][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:50:58,238][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:50:58,735][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:50:59,232][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:50:59,729][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:51:00,227][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:51:00,724][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:51:01,222][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:51:01,719][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:51:02,217][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:51:02,715][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:51:03,213][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:51:03,712][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:51:04,210][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:51:04,708][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:51:05,206][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:51:05,706][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:51:06,206][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:51:06,706][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:51:07,205][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:51:07,705][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:51:08,204][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:51:08,704][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:51:09,202][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:51:09,700][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:51:10,197][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:51:10,695][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:51:11,193][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:51:11,692][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:51:12,192][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:51:12,690][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:51:13,188][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:51:13,687][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:51:14,187][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:51:14,686][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:51:15,185][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:51:15,684][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:51:16,184][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:51:16,682][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:51:17,180][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:51:17,678][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:51:18,178][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:51:18,677][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:51:19,175][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:51:19,674][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:51:20,173][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:51:20,671][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:51:21,170][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:51:21,668][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:51:22,167][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:51:22,666][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:51:23,166][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:51:23,665][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:51:24,165][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:51:24,663][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:51:25,161][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:51:25,659][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:51:26,157][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:51:26,655][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:51:27,154][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:51:27,653][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:51:28,153][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:51:28,651][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:51:29,148][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:51:29,644][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:51:30,141][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:51:30,637][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:51:31,134][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:51:31,631][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:51:32,130][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:51:32,628][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 01:51:33,764][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 01:51:34,824][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:51:34,826][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:51:34,828][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:51:35,533][__main__][INFO] - Iteration 247 took 1m 14s (8.21% Gen, 90.85% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 53m 36s. Estimated total time: 62h 10m 30s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 21s, 500 more iterations: 10h 21m 45s. [2026-03-26 01:51:35,535][__main__][INFO] - Starting iteration 247. [2026-03-26 01:51:35,935][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:51:35,936][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:51:42,189][__main__][INFO] - Number of regex retries in iteration 247: 0 [2026-03-26 01:51:42,189][__main__][INFO] - agents played in iteration 247 are Alice, Bob [2026-03-26 01:51:43,061][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:51:43,105][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:51:43,106][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:51:43,106][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:51:43,710][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:51:44,166][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:51:44,670][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:51:45,170][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:51:45,669][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:51:46,168][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:51:46,667][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:51:47,167][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:51:47,666][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:51:48,166][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:51:48,665][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:51:49,163][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:51:49,660][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:51:50,157][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:51:50,655][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:51:51,152][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:51:51,651][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:51:52,151][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:51:52,652][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:51:53,151][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:51:53,650][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:51:54,149][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:51:54,648][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:51:55,147][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:51:55,647][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:51:56,145][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:51:56,644][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:51:57,142][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:51:57,640][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:51:58,138][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:51:58,636][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:51:59,134][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:51:59,632][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:52:00,133][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:52:00,632][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:52:01,131][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:52:01,630][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:52:02,129][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:52:02,629][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:52:03,129][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:52:03,628][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:52:04,126][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:52:04,625][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:52:05,124][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:52:05,621][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:52:06,119][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:52:06,617][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:52:07,115][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:52:07,613][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:52:08,112][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:52:08,611][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:52:09,110][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:52:09,609][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:52:10,108][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:52:10,608][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:52:11,108][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:52:11,607][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:52:12,107][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:52:12,606][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:52:13,105][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:52:13,603][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:52:14,102][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:52:14,600][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:52:15,099][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:52:15,600][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:52:16,101][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:52:16,602][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:52:17,101][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:52:17,599][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:52:18,098][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:52:18,599][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:52:19,098][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:52:19,597][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:52:20,096][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:52:20,593][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:52:21,091][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:52:21,589][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:52:22,088][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:52:22,586][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:52:23,085][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:52:23,584][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:52:24,084][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:52:24,584][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:52:25,083][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:52:25,582][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:52:26,081][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:52:26,581][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:52:27,081][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:52:27,579][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:52:28,078][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:52:28,575][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:52:29,073][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:52:29,570][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:52:30,069][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:52:30,567][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:52:31,065][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:52:31,564][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:52:32,064][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:52:32,564][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:52:33,063][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:52:33,563][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:52:34,062][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:52:34,562][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:52:35,061][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:52:35,560][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:52:36,058][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:52:36,557][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:52:37,054][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:52:37,552][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:52:38,049][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:52:38,547][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:52:39,045][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:52:39,544][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:52:40,043][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:52:40,541][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:52:41,040][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:52:41,538][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:52:42,036][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:52:42,535][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:52:43,034][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:52:43,533][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:52:44,032][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:52:44,529][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:52:45,026][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:52:45,523][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:52:46,020][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:52:46,517][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:52:47,014][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:52:47,512][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 01:52:48,141][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 01:52:48,874][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:52:48,876][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:52:48,878][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:52:49,527][__main__][INFO] - Iteration 248 took 1m 13s (8.50% Gen, 90.62% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 1m 28s. Estimated total time: 61h 19m 35s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 39s, 500 more iterations: 10h 13m 15s. [2026-03-26 01:52:49,529][__main__][INFO] - Starting iteration 248. [2026-03-26 01:52:49,930][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:52:49,931][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:52:52,633][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:52:53,926][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:52:54,421][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:52:56,624][__main__][INFO] - Number of regex retries in iteration 248: 3 [2026-03-26 01:52:56,624][__main__][INFO] - agents played in iteration 248 are Alice, Bob [2026-03-26 01:52:57,503][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:52:57,548][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:52:57,548][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:52:57,549][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:52:58,154][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:52:58,610][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:52:59,113][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:52:59,612][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:53:00,111][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:53:00,610][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:53:01,109][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:53:01,608][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:53:02,108][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:53:02,609][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:53:03,109][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:53:03,610][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:53:04,110][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:53:04,611][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:53:05,110][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:53:05,611][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:53:06,111][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:53:06,611][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:53:07,111][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:53:07,611][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:53:08,110][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:53:08,610][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:53:09,110][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:53:09,610][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:53:10,108][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:53:10,607][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:53:11,107][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:53:11,606][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:53:12,105][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:53:12,605][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:53:13,105][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:53:13,604][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:53:14,105][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:53:14,605][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:53:15,105][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:53:15,605][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:53:16,105][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:53:16,604][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:53:17,105][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:53:17,605][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:53:18,107][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:53:18,609][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:53:19,109][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:53:19,609][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:53:20,110][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:53:20,610][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:53:21,109][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:53:21,609][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:53:22,109][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:53:22,608][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:53:23,106][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:53:23,605][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:53:24,102][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:53:24,600][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:53:25,099][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:53:25,597][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:53:26,095][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:53:26,593][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:53:27,091][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:53:27,588][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:53:28,085][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:53:28,583][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:53:29,080][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:53:29,576][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:53:30,073][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:53:30,571][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:53:31,070][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:53:31,568][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:53:32,066][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:53:32,565][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:53:33,062][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:53:33,560][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:53:34,059][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:53:34,556][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:53:35,055][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:53:35,553][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:53:36,052][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:53:36,551][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:53:37,049][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:53:37,548][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:53:38,048][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:53:38,547][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:53:39,047][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:53:39,546][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:53:40,045][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:53:40,544][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:53:41,043][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:53:41,541][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:53:42,040][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:53:42,539][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:53:43,039][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:53:43,539][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:53:44,038][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:53:44,536][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:53:45,036][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:53:45,539][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:53:46,041][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:53:46,540][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:53:47,039][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:53:47,538][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:53:48,038][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:53:48,539][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:53:49,038][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:53:49,536][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:53:50,035][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:53:50,535][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:53:51,035][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:53:51,534][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:53:52,034][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:53:52,534][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:53:53,035][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:53:53,534][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:53:54,034][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:53:54,533][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:53:55,033][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:53:55,531][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:53:56,030][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:53:56,529][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:53:57,028][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:53:57,527][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:53:58,025][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:53:58,523][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:53:59,021][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:53:59,519][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:54:00,016][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:54:00,513][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:54:01,011][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:54:01,510][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:54:02,008][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 01:54:02,648][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:54:03,382][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:54:03,384][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:54:03,386][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:54:04,085][__main__][INFO] - Iteration 249 took 1m 14s (9.03% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 28m 24s. Estimated total time: 61h 47m 46s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 57s. [2026-03-26 01:54:04,088][__main__][INFO] - Starting iteration 249. [2026-03-26 01:54:04,488][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:54:04,489][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:54:05,067][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:54:11,015][__main__][INFO] - Number of regex retries in iteration 249: 1 [2026-03-26 01:54:11,016][__main__][INFO] - agents played in iteration 249 are Alice, Bob [2026-03-26 01:54:11,896][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:54:11,940][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:54:11,941][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:54:11,941][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:54:12,544][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:54:12,998][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:54:13,502][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:54:14,000][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:54:14,499][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:54:14,996][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:54:15,494][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:54:15,993][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:54:16,493][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:54:16,994][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:54:17,492][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:54:17,992][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:54:18,492][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:54:18,991][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:54:19,491][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:54:19,991][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:54:20,490][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:54:20,992][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:54:21,492][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:54:21,992][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:54:22,492][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:54:22,993][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:54:23,494][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:54:23,994][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:54:24,494][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:54:24,993][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:54:25,491][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:54:25,990][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:54:26,490][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:54:26,988][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:54:27,486][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:54:27,985][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:54:28,484][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:54:28,983][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:54:29,481][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:54:29,979][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:54:30,477][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:54:30,976][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:54:31,475][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:54:31,974][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:54:32,473][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:54:32,971][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:54:33,469][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:54:33,967][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:54:34,466][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:54:34,966][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:54:35,465][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:54:35,964][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:54:36,463][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:54:36,961][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:54:37,458][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:54:37,955][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:54:38,453][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:54:38,951][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:54:39,449][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:54:39,947][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:54:40,445][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:54:40,944][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:54:41,442][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:54:41,940][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:54:42,438][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:54:42,936][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:54:43,434][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:54:43,934][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:54:44,433][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:54:44,931][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:54:45,430][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:54:45,928][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:54:46,425][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:54:46,923][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:54:47,422][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:54:47,921][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:54:48,421][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:54:48,920][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:54:49,419][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:54:49,917][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:54:50,416][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:54:50,914][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:54:51,413][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:54:51,912][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:54:52,412][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:54:52,911][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:54:53,411][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:54:53,911][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:54:54,411][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:54:54,911][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:54:55,412][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:54:55,913][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:54:56,412][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:54:56,910][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:54:57,409][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:54:57,908][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:54:58,407][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:54:58,906][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:54:59,405][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:54:59,903][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:55:00,402][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:55:00,900][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:55:01,398][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:55:01,897][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:55:02,395][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:55:02,894][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:55:03,394][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:55:03,893][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:55:04,392][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:55:04,890][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:55:05,387][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:55:05,886][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:55:06,384][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:55:06,881][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:55:07,379][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:55:07,876][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:55:08,373][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:55:08,870][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:55:09,368][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:55:09,865][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:55:10,361][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:55:10,858][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:55:11,355][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:55:11,853][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:55:12,350][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:55:12,848][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:55:13,346][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:55:13,844][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:55:14,341][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:55:14,839][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:55:15,338][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:55:15,839][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:55:16,339][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 01:55:16,969][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:55:17,709][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:55:17,711][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:55:17,713][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:55:18,362][__main__][INFO] - Iteration 250 took 1m 13s (8.84% Gen, 90.28% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 13m 7s. Estimated total time: 61h 33m 43s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 7s, 500 more iterations: 10h 15m 37s. [2026-03-26 01:55:18,364][__main__][INFO] - Starting iteration 250. [2026-03-26 01:55:18,764][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 4 and human policies 1. [2026-03-26 01:55:18,765][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:55:25,090][__main__][INFO] - Number of regex retries in iteration 250: 0 [2026-03-26 01:55:25,091][__main__][INFO] - agents played in iteration 250 are Alice, Bob [2026-03-26 01:55:25,968][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:55:26,288][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:55:26,288][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:55:26,289][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:55:26,891][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:55:27,344][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:55:27,846][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:55:28,344][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:55:28,841][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:55:29,340][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:55:29,837][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:55:30,335][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:55:30,833][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:55:31,331][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:55:31,829][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:55:32,327][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:55:32,825][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:55:33,323][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:55:33,819][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:55:34,317][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:55:34,815][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:55:35,316][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:55:35,816][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:55:36,315][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:55:36,814][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:55:37,314][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:55:37,813][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:55:38,313][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:55:38,812][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:55:39,311][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:55:39,810][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:55:40,308][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:55:40,806][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:55:41,305][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:55:41,803][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:55:42,301][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:55:42,798][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:55:43,298][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:55:43,798][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:55:44,296][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:55:44,794][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:55:45,292][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:55:45,791][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:55:46,291][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:55:46,789][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:55:47,287][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:55:47,786][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:55:48,287][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:55:48,786][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:55:49,285][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:55:49,783][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:55:50,281][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:55:50,780][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:55:51,282][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:55:51,783][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:55:52,282][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:55:52,782][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:55:53,282][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:55:53,782][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:55:54,283][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:55:54,783][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:55:55,283][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:55:55,782][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:55:56,281][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:55:56,780][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:55:57,279][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:55:57,777][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:55:58,277][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:55:58,777][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:55:59,275][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:55:59,773][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:56:00,271][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:56:00,768][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:56:01,267][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:56:01,765][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:56:02,265][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:56:02,763][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:56:03,261][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:56:03,759][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:56:04,257][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:56:04,754][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:56:05,251][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:56:05,749][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:56:06,247][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:56:06,746][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:56:07,244][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:56:07,743][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:56:08,240][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:56:08,739][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:56:09,237][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:56:09,735][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:56:10,234][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:56:10,732][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:56:11,230][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:56:11,728][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:56:12,227][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:56:12,724][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:56:13,221][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:56:13,719][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:56:14,216][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:56:14,714][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:56:15,214][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:56:15,713][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:56:16,212][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:56:16,711][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:56:17,210][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:56:17,710][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:56:18,211][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:56:18,712][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:56:19,211][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:56:19,710][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:56:20,209][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:56:20,707][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:56:21,206][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:56:21,707][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:56:22,206][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:56:22,705][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:56:23,205][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:56:23,706][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:56:24,207][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:56:24,708][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:56:25,209][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:56:25,710][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:56:26,210][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:56:26,711][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:56:27,210][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:56:27,711][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:56:28,211][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:56:28,711][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:56:29,212][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:56:29,713][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:56:30,213][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:56:30,712][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 01:56:31,384][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 01:56:32,124][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:56:32,126][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:56:32,128][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:56:33,404][__main__][INFO] - Iteration 251 took 1m 14s (8.48% Gen, 89.81% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 50m 11s. Estimated total time: 62h 12m 2s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 0s. [2026-03-26 01:56:33,408][__main__][INFO] - Starting iteration 251. [2026-03-26 01:56:33,815][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 01:56:33,815][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:56:34,402][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:56:40,618][__main__][INFO] - Number of regex retries in iteration 251: 1 [2026-03-26 01:56:40,619][__main__][INFO] - agents played in iteration 251 are Alice, Bob [2026-03-26 01:56:41,494][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:56:41,538][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:56:41,539][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:56:41,540][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:56:42,145][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:56:42,601][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:56:43,104][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:56:43,604][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:56:44,103][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:56:44,601][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:56:45,100][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:56:45,601][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:56:46,101][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:56:46,599][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:56:47,097][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:56:47,595][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:56:48,094][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:56:48,594][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:56:49,093][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:56:49,592][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:56:50,091][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:56:50,589][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:56:51,087][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:56:51,586][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:56:52,083][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:56:52,581][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:56:53,078][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:56:53,576][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:56:54,074][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:56:54,574][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:56:55,074][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:56:55,573][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:56:56,072][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:56:56,571][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:56:57,070][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:56:57,569][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:56:58,068][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:56:58,565][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:56:59,063][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:56:59,562][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:57:00,060][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:57:00,557][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:57:01,055][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:57:01,554][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:57:02,052][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:57:02,551][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:57:03,050][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:57:03,548][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:57:04,046][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:57:04,545][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:57:05,045][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:57:05,545][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:57:06,044][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:57:06,544][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:57:07,044][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:57:07,543][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:57:08,042][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:57:08,541][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:57:09,040][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:57:09,540][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:57:10,039][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:57:10,539][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:57:11,038][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:57:11,537][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:57:12,036][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:57:12,535][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:57:13,035][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:57:13,535][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:57:14,034][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:57:14,533][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:57:15,032][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:57:15,532][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:57:16,031][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:57:16,530][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:57:17,028][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:57:17,527][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:57:18,027][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:57:18,527][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:57:19,025][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:57:19,523][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:57:20,021][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:57:20,519][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:57:21,018][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:57:21,517][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:57:22,015][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:57:22,514][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:57:23,013][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:57:23,511][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:57:24,011][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:57:24,510][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:57:25,009][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:57:25,509][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:57:26,008][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:57:26,507][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:57:27,007][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:57:27,506][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:57:28,006][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:57:28,505][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:57:29,004][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:57:29,503][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:57:30,003][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:57:30,500][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:57:30,998][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:57:31,496][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:57:31,994][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:57:32,492][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:57:32,991][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:57:33,489][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:57:33,987][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:57:34,484][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:57:34,982][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:57:35,480][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:57:35,976][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:57:36,475][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:57:36,973][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:57:37,472][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:57:37,970][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:57:38,469][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:57:38,970][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:57:39,469][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:57:39,969][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:57:40,468][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:57:40,968][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:57:41,468][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:57:41,967][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:57:42,466][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:57:42,963][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:57:43,461][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:57:43,958][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:57:44,457][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:57:44,955][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:57:45,454][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:57:45,953][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 01:57:46,597][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 01:57:47,328][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:57:47,331][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:57:47,332][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:57:47,992][__main__][INFO] - Iteration 252 took 1m 14s (9.17% Gen, 89.94% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 25m 47s. Estimated total time: 61h 48m 53s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 37s, 500 more iterations: 10h 18m 8s. [2026-03-26 01:57:47,994][__main__][INFO] - Starting iteration 252. [2026-03-26 01:57:48,396][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 01:57:48,396][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:57:50,638][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:57:54,981][__main__][INFO] - Number of regex retries in iteration 252: 1 [2026-03-26 01:57:54,981][__main__][INFO] - agents played in iteration 252 are Alice, Bob [2026-03-26 01:57:55,853][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:57:55,897][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:57:55,898][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:57:55,898][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:57:56,511][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:57:56,966][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:57:57,471][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:57:57,972][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:57:58,472][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:57:58,972][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:57:59,472][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:57:59,972][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:58:00,472][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:58:00,971][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:58:01,469][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:58:01,968][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:58:02,467][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:58:02,966][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:58:03,463][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:58:03,961][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:58:04,459][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:58:04,958][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:58:05,457][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:58:05,955][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:58:06,454][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:58:06,952][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:58:07,450][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:58:07,947][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:58:08,445][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:58:08,944][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:58:09,443][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:58:09,942][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:58:10,440][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:58:10,939][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:58:11,437][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:58:11,935][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:58:12,433][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:58:12,932][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:58:13,430][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:58:13,930][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:58:14,429][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:58:14,928][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:58:15,429][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:58:15,930][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:58:16,429][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:58:16,927][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:58:17,427][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:58:17,927][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:58:18,428][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:58:18,927][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:58:19,424][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:58:19,922][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:58:20,420][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:58:20,917][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:58:21,415][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:58:21,914][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:58:22,411][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:58:22,909][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:58:23,406][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:58:23,903][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:58:24,400][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:58:24,897][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:58:25,394][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:58:25,892][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:58:26,390][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:58:26,888][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:58:27,385][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:58:27,883][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:58:28,380][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:58:28,879][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:58:29,378][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:58:29,876][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:58:30,375][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:58:30,875][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:58:31,374][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:58:31,873][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:58:32,372][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:58:32,870][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:58:33,366][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:58:33,864][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:58:34,361][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:58:34,858][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:58:35,355][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:58:35,854][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:58:36,352][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:58:36,850][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:58:37,347][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:58:37,846][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:58:38,344][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:58:38,842][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:58:39,340][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:58:39,839][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:58:40,338][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:58:40,836][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:58:41,334][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:58:41,832][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:58:42,331][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:58:42,829][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:58:43,328][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:58:43,827][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:58:44,326][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:58:44,821][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:58:45,319][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 01:58:45,818][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 01:58:46,315][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 01:58:46,813][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 01:58:47,313][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 01:58:47,813][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 01:58:48,314][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 01:58:48,813][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 01:58:49,312][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 01:58:49,811][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 01:58:50,310][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 01:58:50,809][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 01:58:51,309][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 01:58:51,808][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 01:58:52,307][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 01:58:52,806][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 01:58:53,303][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 01:58:53,800][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 01:58:54,297][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 01:58:54,793][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 01:58:55,291][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 01:58:55,789][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 01:58:56,287][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 01:58:56,784][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 01:58:57,281][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 01:58:57,778][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 01:58:58,276][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 01:58:58,773][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 01:58:59,270][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 01:58:59,769][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 01:59:00,267][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21738 tokens. [2026-03-26 01:59:00,900][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 01:59:01,629][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 01:59:01,632][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 01:59:01,633][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 01:59:02,295][__main__][INFO] - Iteration 253 took 1m 13s (8.91% Gen, 90.19% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 10m 38s. Estimated total time: 61h 34m 58s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 9s, 500 more iterations: 10h 15m 49s. [2026-03-26 01:59:02,297][__main__][INFO] - Starting iteration 253. [2026-03-26 01:59:02,698][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 01:59:02,699][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 01:59:03,298][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:59:04,850][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 01:59:09,186][__main__][INFO] - Number of regex retries in iteration 253: 2 [2026-03-26 01:59:09,187][__main__][INFO] - agents played in iteration 253 are Alice, Bob [2026-03-26 01:59:10,061][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:59:10,105][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 01:59:10,106][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 01:59:10,106][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 01:59:10,718][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 01:59:11,177][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 01:59:11,682][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 01:59:12,182][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 01:59:12,682][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 01:59:13,182][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 01:59:13,681][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 01:59:14,181][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 01:59:14,680][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 01:59:15,178][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 01:59:15,676][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 01:59:16,175][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 01:59:16,673][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 01:59:17,172][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 01:59:17,670][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 01:59:18,171][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 01:59:18,670][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 01:59:19,170][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 01:59:19,668][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 01:59:20,164][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 01:59:20,663][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 01:59:21,160][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 01:59:21,657][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 01:59:22,154][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 01:59:22,652][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 01:59:23,149][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 01:59:23,648][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 01:59:24,147][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 01:59:24,644][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 01:59:25,142][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 01:59:25,640][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 01:59:26,138][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 01:59:26,635][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 01:59:27,133][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 01:59:27,632][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 01:59:28,130][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 01:59:28,629][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 01:59:29,128][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 01:59:29,627][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 01:59:30,125][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 01:59:30,623][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 01:59:31,121][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 01:59:31,618][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 01:59:32,116][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 01:59:32,616][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 01:59:33,113][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 01:59:33,612][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 01:59:34,111][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 01:59:34,610][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 01:59:35,111][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 01:59:35,611][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 01:59:36,112][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 01:59:36,612][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 01:59:37,112][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 01:59:37,612][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 01:59:38,112][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 01:59:38,613][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 01:59:39,113][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 01:59:39,618][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 01:59:40,117][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 01:59:40,616][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 01:59:41,114][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 01:59:41,613][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 01:59:42,111][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 01:59:42,609][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 01:59:43,110][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 01:59:43,611][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 01:59:44,112][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 01:59:44,611][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 01:59:45,112][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 01:59:45,616][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 01:59:46,118][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 01:59:46,619][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 01:59:47,118][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 01:59:47,617][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 01:59:48,116][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 01:59:48,617][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 01:59:49,116][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 01:59:49,615][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 01:59:50,114][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 01:59:50,614][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 01:59:51,113][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 01:59:51,612][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 01:59:52,110][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 01:59:52,610][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 01:59:53,110][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 01:59:53,609][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 01:59:54,109][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 01:59:54,607][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 01:59:55,106][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 01:59:55,604][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 01:59:56,102][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 01:59:56,599][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 01:59:57,096][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 01:59:57,594][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 01:59:58,092][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 01:59:58,590][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 01:59:59,088][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 01:59:59,585][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:00:00,083][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:00:00,581][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:00:01,078][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:00:01,575][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:00:02,073][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:00:02,572][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:00:03,070][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:00:03,567][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:00:04,064][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:00:04,561][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:00:05,058][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:00:05,555][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:00:06,052][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:00:06,550][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:00:07,050][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:00:07,550][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:00:08,051][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:00:08,551][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:00:09,051][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:00:09,551][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:00:10,051][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:00:10,551][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:00:11,050][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:00:11,548][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:00:12,046][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:00:12,544][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:00:13,042][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:00:13,540][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:00:14,038][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:00:14,536][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21738 tokens. [2026-03-26 02:00:15,176][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 02:00:15,922][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:00:15,924][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:00:15,926][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:00:16,572][__main__][INFO] - Iteration 254 took 1m 13s (8.78% Gen, 90.34% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 8m 9s. Estimated total time: 61h 33m 44s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 7s, 500 more iterations: 10h 15m 37s. [2026-03-26 02:00:16,575][__main__][INFO] - Starting iteration 254. [2026-03-26 02:00:16,971][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:00:16,971][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:00:19,091][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:00:23,658][__main__][INFO] - Number of regex retries in iteration 254: 1 [2026-03-26 02:00:23,659][__main__][INFO] - agents played in iteration 254 are Alice, Bob [2026-03-26 02:00:24,536][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:00:24,580][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:00:24,581][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:00:24,582][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:00:25,195][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:00:25,649][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:00:26,151][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:00:26,649][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:00:27,148][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:00:27,647][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:00:28,146][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:00:28,644][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:00:29,142][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:00:29,642][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:00:30,142][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:00:30,642][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:00:31,143][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:00:31,642][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:00:32,142][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:00:32,642][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:00:33,142][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:00:33,640][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:00:34,137][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:00:34,635][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:00:35,134][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:00:35,632][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:00:36,130][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:00:36,628][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:00:37,125][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:00:37,623][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:00:38,122][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:00:38,620][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:00:39,119][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:00:39,619][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:00:40,117][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:00:40,616][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:00:41,115][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:00:41,616][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:00:42,117][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:00:42,618][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:00:43,220][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:00:43,719][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:00:44,220][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:00:44,719][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:00:45,220][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:00:45,719][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:00:46,217][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:00:46,715][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:00:47,213][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:00:47,712][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:00:48,212][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:00:48,711][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:00:49,210][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:00:49,709][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:00:50,207][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:00:50,705][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:00:51,203][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:00:51,701][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:00:52,199][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:00:52,696][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:00:53,196][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:00:53,694][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:00:54,192][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:00:54,691][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:00:55,191][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:00:55,690][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:00:56,189][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:00:56,688][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:00:57,188][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:00:57,687][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:00:58,185][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:00:58,683][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:00:59,181][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:00:59,678][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:01:00,176][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:01:00,674][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:01:01,173][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:01:01,673][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:01:02,174][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:01:02,674][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:01:03,174][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:01:03,675][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:01:04,175][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:01:04,675][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:01:05,175][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:01:05,673][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:01:06,171][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:01:06,668][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:01:07,166][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:01:07,664][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:01:08,162][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:01:08,660][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:01:09,158][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:01:09,657][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:01:10,155][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:01:10,654][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:01:11,153][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:01:11,652][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:01:12,152][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:01:12,651][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:01:13,150][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:01:13,650][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:01:14,150][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:01:14,649][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:01:15,150][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:01:15,654][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:01:16,156][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:01:16,657][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:01:17,157][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:01:17,658][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:01:18,165][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:01:18,665][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:01:19,164][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:01:19,663][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:01:20,161][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:01:20,659][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:01:21,158][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:01:21,656][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:01:22,154][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:01:22,653][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:01:23,152][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:01:23,650][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:01:24,148][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:01:24,646][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:01:25,144][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:01:25,643][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:01:26,142][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:01:26,640][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:01:27,138][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:01:27,637][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:01:28,136][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:01:28,634][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:01:29,133][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21734 tokens. [2026-03-26 02:01:29,779][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:01:30,549][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:01:30,551][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:01:30,553][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:01:31,269][__main__][INFO] - Iteration 255 took 1m 14s (9.00% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 28m 6s. Estimated total time: 61h 54m 55s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 49s, 500 more iterations: 10h 19m 9s. [2026-03-26 02:01:31,271][__main__][INFO] - Starting iteration 255. [2026-03-26 02:01:31,674][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:01:31,675][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:01:34,529][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:01:38,690][__main__][INFO] - Number of regex retries in iteration 255: 1 [2026-03-26 02:01:38,690][__main__][INFO] - agents played in iteration 255 are Alice, Bob [2026-03-26 02:01:39,577][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:01:39,622][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:01:39,622][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:01:39,623][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:01:40,233][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:01:40,686][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:01:41,188][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:01:41,690][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:01:42,188][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:01:42,688][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:01:43,186][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:01:43,683][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:01:44,181][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:01:44,681][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:01:45,179][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:01:45,677][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:01:46,176][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:01:46,675][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:01:47,174][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:01:47,673][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:01:48,173][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:01:48,676][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:01:49,175][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:01:49,676][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:01:50,175][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:01:50,675][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:01:51,175][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:01:51,674][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:01:52,174][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:01:52,674][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:01:53,176][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:01:53,675][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:01:54,175][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:01:54,675][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:01:55,175][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:01:55,674][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:01:56,175][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:01:56,674][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:01:57,172][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:01:57,670][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:01:58,168][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:01:58,665][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:01:59,162][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:01:59,660][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:02:00,158][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:02:00,655][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:02:01,153][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:02:01,650][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:02:02,148][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:02:02,647][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:02:03,144][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:02:03,641][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:02:04,139][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:02:04,656][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:02:05,159][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:02:05,659][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:02:06,159][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:02:06,659][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:02:07,160][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:02:07,660][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:02:08,160][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:02:08,658][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:02:09,155][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:02:09,651][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:02:10,148][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:02:10,644][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:02:11,141][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:02:11,638][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:02:12,136][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:02:12,634][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:02:13,131][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:02:13,629][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:02:14,128][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:02:14,626][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:02:15,123][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:02:15,620][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:02:16,118][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:02:16,616][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:02:17,115][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:02:17,614][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:02:18,113][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:02:18,613][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:02:19,111][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:02:19,611][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:02:20,110][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:02:20,610][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:02:21,110][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:02:21,610][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:02:22,110][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:02:22,609][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:02:23,109][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:02:23,610][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:02:24,110][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:02:24,611][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:02:25,110][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:02:25,609][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:02:26,108][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:02:26,609][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:02:27,109][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:02:27,609][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:02:28,110][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:02:28,608][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:02:29,106][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:02:29,603][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:02:30,101][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:02:30,598][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:02:31,096][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:02:31,594][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:02:32,093][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:02:32,590][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:02:33,089][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:02:33,586][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:02:34,083][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:02:34,580][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:02:35,078][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:02:35,576][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:02:36,074][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:02:36,576][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:02:37,076][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:02:37,576][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:02:38,077][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:02:38,577][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:02:39,076][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:02:39,576][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:02:40,077][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:02:40,574][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:02:41,072][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:02:41,570][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:02:42,068][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:02:42,565][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:02:43,063][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:02:43,561][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:02:44,060][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 02:02:44,702][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 02:02:45,459][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:02:45,461][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:02:45,463][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:02:46,343][__main__][INFO] - Iteration 256 took 1m 14s (9.40% Gen, 89.42% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 56h 45m 26s. Estimated total time: 62h 13m 30s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 27s, 500 more iterations: 10h 22m 15s. [2026-03-26 02:02:46,346][__main__][INFO] - Starting iteration 256. [2026-03-26 02:02:46,753][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:02:46,754][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:02:53,229][__main__][INFO] - Number of regex retries in iteration 256: 0 [2026-03-26 02:02:53,230][__main__][INFO] - agents played in iteration 256 are Alice, Bob [2026-03-26 02:02:54,104][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:02:54,149][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:02:54,149][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:02:54,150][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:02:54,757][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:02:55,213][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:02:55,717][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:02:56,216][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:02:56,716][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:02:57,215][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:02:57,716][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:02:58,216][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:02:58,715][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:02:59,217][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:02:59,717][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:03:00,217][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:03:00,717][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:03:01,217][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:03:01,717][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:03:02,218][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:03:02,717][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:03:03,216][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:03:03,716][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:03:04,216][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:03:04,715][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:03:05,214][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:03:05,713][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:03:06,212][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:03:06,711][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:03:07,210][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:03:07,709][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:03:08,208][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:03:08,706][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:03:09,203][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:03:09,701][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:03:10,199][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:03:10,696][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:03:11,195][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:03:11,695][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:03:12,195][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:03:12,696][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:03:13,195][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:03:13,695][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:03:14,195][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:03:14,695][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:03:15,194][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:03:15,692][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:03:16,191][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:03:16,689][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:03:17,188][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:03:17,687][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:03:18,187][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:03:18,686][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:03:19,184][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:03:19,682][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:03:20,180][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:03:20,677][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:03:21,175][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:03:21,674][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:03:22,171][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:03:22,669][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:03:23,167][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:03:23,665][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:03:24,163][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:03:24,661][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:03:25,160][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:03:25,844][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:03:26,395][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:03:26,894][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:03:27,393][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:03:27,892][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:03:28,391][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:03:28,890][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:03:29,390][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:03:29,890][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:03:30,389][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:03:30,890][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:03:31,390][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:03:31,889][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:03:32,388][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:03:32,886][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:03:33,385][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:03:33,884][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:03:34,382][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:03:34,881][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:03:35,380][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:03:35,878][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:03:36,508][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:03:37,007][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:03:37,505][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:03:38,004][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:03:38,503][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:03:39,001][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:03:39,499][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:03:39,997][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:03:40,495][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:03:40,993][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:03:41,491][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:03:41,989][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:03:42,486][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:03:42,983][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:03:43,483][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:03:43,983][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:03:44,483][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:03:44,982][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:03:45,481][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:03:45,981][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:03:46,479][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:03:46,978][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:03:47,477][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:03:47,976][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:03:48,477][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:03:48,975][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:03:49,473][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:03:49,972][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:03:50,471][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:03:50,970][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:03:51,469][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:03:51,968][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:03:52,466][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:03:52,966][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:03:53,463][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:03:53,961][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:03:54,459][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:03:54,957][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:03:55,456][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:03:55,955][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:03:56,454][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:03:56,952][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:03:57,450][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:03:57,948][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:03:58,446][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:03:58,944][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:03:59,582][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:04:00,240][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:04:00,242][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:04:00,244][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:04:00,902][__main__][INFO] - Iteration 257 took 1m 14s (8.73% Gen, 90.38% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 18m 11s. Estimated total time: 61h 47m 30s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 55s. [2026-03-26 02:04:00,906][__main__][INFO] - Starting iteration 257. [2026-03-26 02:04:01,306][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:04:01,306][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:04:06,818][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:04:08,400][__main__][INFO] - Number of regex retries in iteration 257: 1 [2026-03-26 02:04:08,400][__main__][INFO] - agents played in iteration 257 are Alice, Bob [2026-03-26 02:04:09,284][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:04:09,328][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:04:09,329][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:04:09,329][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:04:09,932][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:04:10,386][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:04:10,888][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:04:11,386][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:04:11,885][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:04:12,383][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:04:12,881][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:04:13,379][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:04:13,877][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:04:14,377][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:04:14,876][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:04:15,377][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:04:15,877][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:04:16,376][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:04:16,875][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:04:17,373][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:04:17,873][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:04:18,375][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:04:18,875][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:04:19,374][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:04:19,874][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:04:20,373][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:04:20,873][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:04:21,372][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:04:21,871][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:04:22,369][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:04:22,867][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:04:23,364][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:04:23,862][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:04:24,359][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:04:24,857][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:04:25,355][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:04:25,853][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:04:26,352][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:04:26,852][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:04:27,350][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:04:27,850][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:04:28,348][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:04:28,849][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:04:29,350][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:04:29,850][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:04:30,348][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:04:30,846][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:04:31,345][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:04:31,843][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:04:32,341][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:04:32,838][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:04:33,336][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:04:33,835][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:04:34,334][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:04:34,833][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:04:35,332][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:04:35,831][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:04:36,330][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:04:36,831][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:04:37,330][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:04:37,830][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:04:38,330][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:04:38,829][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:04:39,327][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:04:39,825][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:04:40,324][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:04:40,823][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:04:41,321][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:04:41,821][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:04:42,319][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:04:42,818][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:04:43,316][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:04:43,815][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:04:44,313][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:04:44,812][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:04:45,311][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:04:45,810][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:04:46,310][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:04:46,808][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:04:47,307][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:04:47,806][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:04:48,307][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:04:48,808][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:04:49,307][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:04:49,806][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:04:50,304][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:04:50,803][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:04:51,302][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:04:51,801][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:04:52,300][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:04:52,800][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:04:53,300][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:04:53,799][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:04:54,296][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:04:54,795][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:04:55,295][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:04:55,793][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:04:56,292][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:04:56,791][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:04:57,290][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:04:57,790][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:04:58,288][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:04:58,787][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:04:59,285][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:04:59,784][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:05:00,282][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:05:00,780][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:05:01,279][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:05:01,778][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:05:02,275][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:05:02,774][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:05:03,272][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:05:03,771][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:05:04,270][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:05:04,770][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:05:05,268][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:05:05,766][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:05:06,264][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:05:06,764][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:05:07,262][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:05:07,760][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:05:08,259][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:05:08,759][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:05:09,258][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:05:09,757][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:05:10,256][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:05:10,755][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:05:11,253][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:05:11,752][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:05:12,251][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:05:12,750][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:05:13,250][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:05:13,749][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 02:05:14,420][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 02:05:15,162][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:05:15,165][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:05:15,166][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:05:15,779][__main__][INFO] - Iteration 258 took 1m 14s (9.53% Gen, 89.65% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 56h 33m 8s. Estimated total time: 62h 3m 42s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 7s, 500 more iterations: 10h 20m 37s. [2026-03-26 02:05:15,782][__main__][INFO] - Starting iteration 258. [2026-03-26 02:05:16,185][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:05:16,186][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:05:18,346][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:05:22,683][__main__][INFO] - Number of regex retries in iteration 258: 1 [2026-03-26 02:05:22,684][__main__][INFO] - agents played in iteration 258 are Alice, Bob [2026-03-26 02:05:23,558][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:05:23,602][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:05:23,603][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:05:23,604][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:05:24,215][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:05:24,674][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:05:25,177][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:05:25,674][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:05:26,173][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:05:26,672][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:05:27,170][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:05:27,668][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:05:28,167][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:05:28,667][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:05:29,166][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:05:29,664][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:05:30,162][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:05:30,661][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:05:31,159][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:05:31,658][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:05:32,157][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:05:32,656][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:05:33,154][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:05:33,652][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:05:34,152][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:05:34,650][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:05:35,150][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:05:35,648][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:05:36,147][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:05:36,645][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:05:37,143][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:05:37,641][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:05:38,139][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:05:38,637][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:05:39,135][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:05:39,633][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:05:40,131][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:05:40,632][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:05:41,131][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:05:41,630][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:05:42,128][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:05:42,630][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:05:43,130][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:05:43,630][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:05:44,129][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:05:44,627][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:05:45,126][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:05:45,627][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:05:46,129][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:05:46,630][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:05:47,128][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:05:47,627][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:05:48,126][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:05:48,627][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:05:49,127][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:05:49,627][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:05:50,126][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:05:50,625][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:05:51,125][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:05:51,624][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:05:52,123][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:05:52,622][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:05:53,120][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:05:53,619][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:05:54,119][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:05:54,619][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:05:55,118][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:05:55,617][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:05:56,116][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:05:56,614][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:05:57,112][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:05:57,612][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:05:58,110][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:05:58,608][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:05:59,107][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:05:59,604][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:06:00,102][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:06:00,599][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:06:01,097][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:06:01,595][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:06:02,093][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:06:02,592][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:06:03,090][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:06:03,589][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:06:04,086][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:06:04,584][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:06:05,082][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:06:05,580][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:06:06,078][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:06:06,577][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:06:07,077][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:06:07,575][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:06:08,074][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:06:08,572][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:06:09,070][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:06:09,567][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:06:10,065][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:06:10,563][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:06:11,061][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:06:11,559][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:06:12,058][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:06:12,557][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:06:13,056][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:06:13,555][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:06:14,054][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:06:14,555][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:06:15,054][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:06:15,553][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:06:16,052][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:06:16,550][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:06:17,048][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:06:17,546][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:06:18,045][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:06:18,546][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:06:19,044][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:06:19,543][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:06:20,041][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:06:20,540][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:06:21,039][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:06:21,539][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:06:22,039][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:06:22,540][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:06:23,039][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:06:23,539][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:06:24,038][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:06:24,537][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:06:25,036][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:06:25,536][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:06:26,035][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:06:26,534][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:06:27,034][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:06:27,533][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:06:28,031][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 02:06:28,671][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 02:06:29,424][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:06:29,426][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:06:29,428][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:06:30,082][__main__][INFO] - Iteration 259 took 1m 13s (8.79% Gen, 90.32% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 3m 3s. Estimated total time: 61h 34m 51s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 9s, 500 more iterations: 10h 15m 48s. [2026-03-26 02:06:30,084][__main__][INFO] - Starting iteration 259. [2026-03-26 02:06:30,485][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:06:30,486][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:06:36,835][__main__][INFO] - Number of regex retries in iteration 259: 0 [2026-03-26 02:06:36,836][__main__][INFO] - agents played in iteration 259 are Alice, Bob [2026-03-26 02:06:37,712][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:06:37,757][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:06:37,757][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:06:37,758][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:06:38,362][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:06:38,817][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:06:39,319][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:06:39,819][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:06:40,318][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:06:40,817][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:06:41,316][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:06:41,815][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:06:42,315][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:06:42,815][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:06:43,316][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:06:43,816][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:06:44,316][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:06:44,816][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:06:45,315][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:06:45,815][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:06:46,314][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:06:46,814][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:06:47,313][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:06:47,814][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:06:48,315][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:06:48,815][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:06:49,315][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:06:49,815][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:06:50,314][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:06:50,814][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:06:51,313][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:06:51,813][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:06:52,313][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:06:52,812][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:06:53,313][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:06:53,813][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:06:54,313][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:06:54,813][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:06:55,313][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:06:55,812][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:06:56,310][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:06:56,810][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:06:57,309][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:06:57,811][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:06:58,311][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:06:58,810][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:06:59,309][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:06:59,807][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:07:00,305][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:07:00,801][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:07:01,299][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:07:01,797][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:07:02,296][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:07:02,794][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:07:03,292][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:07:03,789][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:07:04,286][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:07:04,783][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:07:05,280][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:07:05,777][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:07:06,274][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:07:06,776][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:07:07,275][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:07:07,775][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:07:08,274][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:07:08,774][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:07:09,273][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:07:09,774][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:07:10,274][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:07:10,772][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:07:11,270][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:07:11,770][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:07:12,268][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:07:12,766][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:07:13,264][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:07:13,763][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:07:14,261][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:07:14,760][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:07:15,261][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:07:15,763][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:07:16,263][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:07:16,762][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:07:17,262][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:07:17,761][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:07:18,262][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:07:18,762][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:07:19,262][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:07:19,761][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:07:20,260][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:07:20,759][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:07:21,259][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:07:21,758][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:07:22,258][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:07:22,757][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:07:23,256][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:07:23,756][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:07:24,255][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:07:24,755][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:07:25,256][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:07:25,757][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:07:26,257][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:07:26,756][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:07:27,254][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:07:27,754][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:07:28,253][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:07:28,753][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:07:29,253][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:07:29,752][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:07:30,251][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:07:30,749][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:07:31,248][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:07:31,746][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:07:32,244][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:07:32,743][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:07:33,241][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:07:33,740][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:07:34,239][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:07:34,737][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:07:35,236][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:07:35,734][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:07:36,231][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:07:36,731][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:07:37,229][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:07:37,727][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:07:38,225][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:07:38,725][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:07:39,224][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:07:39,724][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:07:40,223][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:07:40,723][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:07:41,223][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:07:41,723][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:07:42,223][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 02:07:42,864][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 02:07:43,621][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:07:43,623][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:07:43,625][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:07:44,284][__main__][INFO] - Iteration 260 took 1m 13s (8.60% Gen, 90.50% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 56m 54s. Estimated total time: 61h 29m 56s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 59s, 500 more iterations: 10h 14m 59s. [2026-03-26 02:07:44,286][__main__][INFO] - Starting iteration 260. [2026-03-26 02:07:44,692][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:07:44,693][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:07:51,295][__main__][INFO] - Number of regex retries in iteration 260: 0 [2026-03-26 02:07:51,296][__main__][INFO] - agents played in iteration 260 are Alice, Bob [2026-03-26 02:07:52,185][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:07:52,230][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:07:52,230][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:07:52,231][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:07:52,831][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:07:53,287][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:07:53,792][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:07:54,292][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:07:54,793][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:07:55,295][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:07:55,795][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:07:56,296][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:07:56,795][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:07:57,295][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:07:57,795][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:07:58,294][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:07:58,794][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:07:59,293][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:07:59,793][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:08:00,292][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:08:00,792][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:08:01,292][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:08:01,792][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:08:02,293][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:08:02,792][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:08:03,292][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:08:03,793][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:08:04,293][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:08:04,792][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:08:05,293][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:08:05,792][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:08:06,292][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:08:06,792][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:08:07,291][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:08:07,791][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:08:08,290][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:08:08,792][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:08:09,291][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:08:09,791][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:08:10,291][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:08:10,792][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:08:11,291][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:08:11,791][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:08:12,293][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:08:12,792][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:08:13,291][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:08:13,788][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:08:14,285][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:08:14,782][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:08:15,279][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:08:15,777][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:08:16,274][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:08:16,773][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:08:17,270][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:08:17,764][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:08:18,262][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:08:18,758][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:08:19,254][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:08:19,749][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:08:20,243][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:08:20,738][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:08:21,236][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:08:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:08:22,234][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:08:22,732][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:08:23,231][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:08:23,729][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:08:24,228][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:08:24,726][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:08:25,226][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:08:25,726][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:08:26,226][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:08:26,726][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:08:27,226][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:08:27,726][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:08:28,227][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:08:28,726][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:08:29,224][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:08:29,723][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:08:30,222][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:08:30,721][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:08:31,220][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:08:31,719][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:08:32,218][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:08:32,717][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:08:33,218][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:08:33,718][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:08:34,217][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:08:34,716][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:08:35,216][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:08:35,715][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:08:36,215][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:08:36,714][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:08:37,214][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:08:37,713][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:08:38,212][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:08:38,712][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:08:39,211][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:08:39,710][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:08:40,209][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:08:40,709][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:08:41,209][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:08:41,708][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:08:42,207][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:08:42,706][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:08:43,205][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:08:43,704][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:08:44,202][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:08:44,700][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:08:45,199][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:08:45,701][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:08:46,200][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:08:46,699][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:08:47,199][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:08:47,697][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:08:48,199][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:08:48,698][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:08:49,197][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:08:49,692][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:08:50,186][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:08:50,681][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:08:51,177][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:08:51,673][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:08:52,168][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:08:52,663][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:08:53,160][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:08:53,658][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:08:54,160][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:08:54,659][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:08:55,158][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:08:55,656][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:08:56,154][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:08:56,652][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 02:08:57,272][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 02:08:58,009][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:08:58,012][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:08:58,013][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:08:58,677][__main__][INFO] - Iteration 261 took 1m 13s (8.92% Gen, 90.18% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 4m 58s. Estimated total time: 61h 39m 15s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 18s, 500 more iterations: 10h 16m 32s. [2026-03-26 02:08:58,679][__main__][INFO] - Starting iteration 261. [2026-03-26 02:08:59,081][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:08:59,082][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:09:05,763][__main__][INFO] - Number of regex retries in iteration 261: 0 [2026-03-26 02:09:05,764][__main__][INFO] - agents played in iteration 261 are Alice, Bob [2026-03-26 02:09:06,657][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:09:06,701][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:09:06,702][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:09:06,703][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:09:07,340][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:09:07,795][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:09:08,298][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:09:08,798][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:09:09,296][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:09:09,795][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:09:10,294][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:09:10,793][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:09:11,292][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:09:11,792][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:09:12,289][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:09:12,786][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:09:13,284][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:09:13,781][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:09:14,278][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:09:14,775][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:09:15,272][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:09:15,771][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:09:16,271][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:09:16,770][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:09:17,268][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:09:17,766][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:09:18,265][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:09:18,765][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:09:19,262][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:09:19,762][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:09:20,260][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:09:20,758][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:09:21,256][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:09:21,755][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:09:22,254][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:09:22,752][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:09:23,250][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:09:23,748][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:09:24,245][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:09:24,743][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:09:25,242][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:09:25,740][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:09:26,239][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:09:26,737][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:09:27,236][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:09:27,735][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:09:28,234][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:09:28,732][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:09:29,230][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:09:29,730][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:09:30,228][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:09:30,726][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:09:31,224][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:09:31,722][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:09:32,220][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:09:32,718][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:09:33,216][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:09:33,714][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:09:34,212][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:09:34,710][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:09:35,208][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:09:35,707][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:09:36,205][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:09:36,703][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:09:37,202][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:09:37,701][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:09:38,201][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:09:38,699][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:09:39,198][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:09:39,697][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:09:40,196][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:09:40,693][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:09:41,192][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:09:41,691][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:09:42,190][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:09:42,688][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:09:43,188][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:09:43,686][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:09:44,183][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:09:44,681][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:09:45,178][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:09:45,675][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:09:46,173][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:09:46,670][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:09:47,168][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:09:47,666][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:09:48,175][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:09:48,677][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:09:49,177][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:09:49,677][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:09:50,176][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:09:50,676][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:09:51,176][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:09:51,677][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:09:52,178][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:09:52,678][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:09:53,180][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:09:53,685][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:09:54,187][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:09:54,688][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:09:55,190][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:09:55,693][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:09:56,193][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:09:56,694][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:09:57,196][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:09:57,699][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:09:58,201][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:09:58,703][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:09:59,204][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:09:59,705][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:10:00,205][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:10:00,704][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:10:01,203][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:10:01,703][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:10:02,202][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:10:02,702][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:10:03,201][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:10:03,700][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:10:04,205][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:10:04,704][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:10:05,202][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:10:05,701][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:10:06,200][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:10:06,699][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:10:07,199][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:10:07,699][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:10:08,199][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:10:08,699][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:10:09,200][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:10:09,700][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:10:10,199][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:10:10,698][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:10:11,198][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 02:10:11,831][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 02:10:12,572][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:10:12,575][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:10:12,576][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:10:13,294][__main__][INFO] - Iteration 262 took 1m 14s (9.00% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 15m 9s. Estimated total time: 61h 50m 40s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 26s. [2026-03-26 02:10:13,297][__main__][INFO] - Starting iteration 262. [2026-03-26 02:10:13,697][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:10:13,697][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:10:14,290][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:10:18,923][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:10:20,415][__main__][INFO] - Number of regex retries in iteration 262: 2 [2026-03-26 02:10:20,416][__main__][INFO] - agents played in iteration 262 are Alice, Bob [2026-03-26 02:10:21,328][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:10:21,372][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:10:21,373][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:10:21,374][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:10:21,978][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:10:22,436][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:10:22,941][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:10:23,441][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:10:23,941][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:10:24,441][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:10:24,941][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:10:25,441][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:10:25,941][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:10:26,441][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:10:26,941][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:10:27,439][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:10:27,938][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:10:28,437][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:10:28,936][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:10:29,434][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:10:29,933][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:10:30,433][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:10:30,931][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:10:31,430][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:10:31,928][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:10:32,427][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:10:32,925][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:10:33,424][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:10:33,924][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:10:34,424][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:10:34,923][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:10:35,423][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:10:35,921][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:10:36,421][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:10:36,921][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:10:37,421][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:10:37,921][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:10:38,420][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:10:38,919][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:10:39,418][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:10:39,917][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:10:40,416][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:10:40,915][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:10:41,413][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:10:41,912][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:10:42,411][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:10:42,910][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:10:43,408][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:10:43,907][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:10:44,406][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:10:44,904][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:10:45,404][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:10:45,903][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:10:46,403][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:10:46,901][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:10:47,400][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:10:47,901][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:10:48,403][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:10:48,903][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:10:49,401][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:10:49,900][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:10:50,398][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:10:50,897][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:10:51,395][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:10:51,894][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:10:52,393][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:10:52,892][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:10:53,390][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:10:53,888][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:10:54,387][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:10:54,886][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:10:55,386][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:10:55,886][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:10:56,385][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:10:56,885][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:10:57,384][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:10:57,884][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:10:58,382][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:10:58,880][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:10:59,378][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:10:59,878][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:11:00,376][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:11:00,875][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:11:01,373][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:11:01,871][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:11:02,369][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:11:02,866][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:11:03,364][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:11:03,861][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:11:04,361][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:11:04,859][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:11:05,357][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:11:05,856][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:11:06,355][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:11:06,855][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:11:07,356][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:11:07,856][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:11:08,356][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:11:08,856][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:11:09,355][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:11:09,854][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:11:10,351][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:11:10,849][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:11:11,347][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:11:11,845][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:11:12,342][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:11:12,840][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:11:13,339][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:11:13,838][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:11:14,336][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:11:14,834][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:11:15,332][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:11:15,832][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:11:16,331][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:11:16,830][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:11:17,329][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:11:17,828][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:11:18,329][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:11:18,827][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:11:19,327][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:11:19,826][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:11:20,325][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:11:20,825][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:11:21,324][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:11:21,823][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:11:22,321][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:11:22,818][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:11:23,318][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:11:23,816][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:11:24,314][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:11:24,812][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:11:25,312][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:11:25,810][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 02:11:26,474][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 02:11:27,210][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:11:27,212][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:11:27,214][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:11:27,866][__main__][INFO] - Iteration 263 took 1m 14s (9.06% Gen, 90.06% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 11m 44s. Estimated total time: 61h 48m 29s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 36s, 500 more iterations: 10h 18m 4s. [2026-03-26 02:11:27,868][__main__][INFO] - Starting iteration 263. [2026-03-26 02:11:28,266][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:11:28,267][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:11:28,862][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:11:30,507][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:11:33,787][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:11:35,252][__main__][INFO] - Number of regex retries in iteration 263: 3 [2026-03-26 02:11:35,253][__main__][INFO] - agents played in iteration 263 are Alice, Bob [2026-03-26 02:11:36,134][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:11:36,178][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:11:36,179][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:11:36,179][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:11:36,794][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:11:37,245][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:11:37,751][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:11:38,252][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:11:38,752][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:11:39,252][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:11:39,751][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:11:40,249][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:11:40,749][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:11:41,250][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:11:41,750][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:11:42,250][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:11:42,749][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:11:43,248][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:11:43,746][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:11:44,245][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:11:44,744][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:11:45,245][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:11:45,748][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:11:46,247][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:11:46,745][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:11:47,243][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:11:47,742][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:11:48,243][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:11:48,742][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:11:49,240][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:11:49,738][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:11:50,236][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:11:50,733][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:11:51,231][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:11:51,729][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:11:52,226][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:11:52,724][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:11:53,226][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:11:53,727][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:11:54,226][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:11:54,726][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:11:55,226][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:11:55,727][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:11:56,227][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:11:56,727][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:11:57,227][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:11:57,724][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:11:58,222][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:11:58,720][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:11:59,218][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:11:59,716][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:12:00,214][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:12:00,713][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:12:01,212][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:12:01,711][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:12:02,209][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:12:02,707][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:12:03,205][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:12:03,703][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:12:04,202][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:12:04,701][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:12:05,201][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:12:05,700][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:12:06,198][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:12:06,698][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:12:07,198][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:12:07,697][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:12:08,196][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:12:08,695][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:12:09,194][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:12:09,692][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:12:10,191][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:12:10,689][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:12:11,187][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:12:11,685][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:12:12,183][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:12:12,682][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:12:13,181][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:12:13,679][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:12:14,178][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:12:14,676][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:12:15,174][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:12:15,673][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:12:16,172][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:12:16,673][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:12:17,172][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:12:17,671][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:12:18,170][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:12:18,670][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:12:19,168][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:12:19,667][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:12:20,164][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:12:20,662][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:12:21,159][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:12:21,658][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:12:22,155][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:12:22,653][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:12:23,150][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:12:23,647][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:12:24,144][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:12:24,641][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:12:25,142][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:12:25,641][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:12:26,142][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:12:26,641][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:12:27,141][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:12:27,641][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:12:28,141][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:12:28,642][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:12:29,140][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:12:29,638][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:12:30,136][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:12:30,633][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:12:31,130][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:12:31,627][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:12:32,125][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:12:32,623][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:12:33,122][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:12:33,622][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:12:34,121][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:12:34,619][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:12:35,118][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:12:35,617][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:12:36,117][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:12:36,616][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:12:37,115][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:12:37,614][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:12:38,114][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:12:38,612][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:12:39,111][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:12:39,610][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:12:40,111][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:12:40,610][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21716 tokens. [2026-03-26 02:12:41,247][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:12:41,977][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:12:41,979][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:12:41,981][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:12:42,633][__main__][INFO] - Iteration 264 took 1m 14s (9.39% Gen, 89.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 20m 23s. Estimated total time: 61h 58m 23s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 56s, 500 more iterations: 10h 19m 43s. [2026-03-26 02:12:42,636][__main__][INFO] - Starting iteration 264. [2026-03-26 02:12:43,036][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:12:43,037][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:12:49,804][__main__][INFO] - Number of regex retries in iteration 264: 0 [2026-03-26 02:12:49,805][__main__][INFO] - agents played in iteration 264 are Alice, Bob [2026-03-26 02:12:50,702][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:12:50,747][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:12:50,747][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:12:50,748][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:12:51,352][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:12:51,808][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:12:52,313][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:12:52,813][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:12:53,313][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:12:53,812][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:12:54,314][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:12:54,815][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:12:55,316][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:12:55,815][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:12:56,314][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:12:56,813][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:12:57,313][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:12:57,814][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:12:58,314][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:12:58,813][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:12:59,312][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:12:59,812][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:13:00,310][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:13:00,808][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:13:01,306][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:13:01,804][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:13:02,302][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:13:02,800][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:13:03,299][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:13:03,798][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:13:04,296][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:13:04,794][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:13:05,292][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:13:05,790][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:13:06,288][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:13:06,787][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:13:07,285][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:13:07,784][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:13:08,282][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:13:08,780][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:13:09,278][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:13:09,775][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:13:10,273][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:13:10,772][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:13:11,271][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:13:11,766][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:13:12,261][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:13:12,756][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:13:13,250][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:13:13,745][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:13:14,240][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:13:14,734][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:13:15,229][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:13:15,729][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:13:16,229][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:13:16,727][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:13:17,225][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:13:17,725][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:13:18,226][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:13:18,725][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:13:19,224][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:13:19,723][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:13:20,222][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:13:20,722][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:13:21,222][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:13:21,722][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:13:22,221][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:13:22,720][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:13:23,219][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:13:23,719][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:13:24,219][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:13:24,718][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:13:25,217][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:13:25,716][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:13:26,215][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:13:26,716][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:13:27,215][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:13:27,715][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:13:28,215][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:13:28,715][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:13:29,213][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:13:29,712][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:13:30,210][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:13:30,709][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:13:31,207][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:13:31,706][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:13:32,204][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:13:32,703][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:13:33,202][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:13:33,700][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:13:34,199][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:13:34,698][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:13:35,197][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:13:35,696][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:13:36,195][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:13:36,694][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:13:37,194][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:13:37,692][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:13:38,190][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:13:38,687][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:13:39,185][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:13:39,680][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:13:40,177][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:13:40,675][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:13:41,173][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:13:41,672][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:13:42,169][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:13:42,668][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:13:43,166][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:13:43,661][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:13:44,156][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:13:44,653][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:13:45,148][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:13:45,642][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:13:46,137][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:13:46,632][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:13:47,127][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:13:47,625][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:13:48,124][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:13:48,622][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:13:49,121][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:13:49,619][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:13:50,117][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:13:50,615][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:13:51,114][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:13:51,614][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:13:52,114][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:13:52,613][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:13:53,113][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:13:53,615][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:13:54,114][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:13:54,614][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:13:55,114][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21723 tokens. [2026-03-26 02:13:55,748][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 02:13:56,613][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:13:56,616][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:13:56,617][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:13:57,587][__main__][INFO] - Iteration 265 took 1m 14s (9.08% Gen, 89.62% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 28m 19s. Estimated total time: 62h 7m 34s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 15s, 500 more iterations: 10h 21m 15s. [2026-03-26 02:13:57,589][__main__][INFO] - Starting iteration 265. [2026-03-26 02:13:57,987][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:13:57,988][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:14:04,408][__main__][INFO] - Number of regex retries in iteration 265: 0 [2026-03-26 02:14:04,409][__main__][INFO] - agents played in iteration 265 are Alice, Bob [2026-03-26 02:14:05,316][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:14:05,360][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:14:05,361][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:14:05,362][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:14:05,960][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:14:06,415][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:14:06,919][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:14:07,419][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:14:07,919][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:14:08,417][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:14:08,917][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:14:09,417][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:14:09,917][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:14:10,413][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:14:10,909][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:14:11,406][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:14:11,903][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:14:12,399][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:14:12,895][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:14:13,392][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:14:13,888][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:14:14,386][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:14:14,883][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:14:15,381][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:14:15,879][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:14:16,376][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:14:16,873][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:14:17,371][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:14:17,867][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:14:18,366][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:14:18,863][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:14:19,363][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:14:19,862][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:14:20,359][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:14:20,857][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:14:21,355][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:14:21,853][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:14:22,350][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:14:22,848][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:14:23,345][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:14:23,844][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:14:24,342][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:14:24,840][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:14:25,339][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:14:25,838][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:14:26,337][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:14:26,835][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:14:27,333][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:14:27,833][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:14:28,332][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:14:28,832][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:14:29,330][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:14:29,830][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:14:30,328][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:14:30,826][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:14:31,323][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:14:31,821][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:14:32,320][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:14:32,818][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:14:33,316][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:14:33,815][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:14:34,314][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:14:34,811][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:14:35,308][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:14:35,805][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:14:36,303][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:14:36,802][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:14:37,300][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:14:37,798][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:14:38,298][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:14:38,795][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:14:39,293][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:14:39,792][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:14:40,291][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:14:40,790][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:14:41,288][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:14:41,786][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:14:42,282][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:14:42,778][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:14:43,274][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:14:43,770][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:14:44,266][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:14:44,762][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:14:45,259][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:14:45,758][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:14:46,257][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:14:46,755][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:14:47,252][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:14:47,750][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:14:48,249][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:14:48,747][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:14:49,245][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:14:49,742][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:14:50,241][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:14:50,739][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:14:51,238][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:14:51,737][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:14:52,235][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:14:52,733][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:14:53,230][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:14:53,728][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:14:54,226][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:14:54,724][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:14:55,223][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:14:55,722][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:14:56,220][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:14:56,719][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:14:57,218][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:14:57,718][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:14:58,216][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:14:58,715][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:14:59,214][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:14:59,713][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:15:00,212][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:15:00,710][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:15:01,207][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:15:01,706][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:15:02,204][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:15:02,702][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:15:03,199][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:15:03,699][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:15:04,194][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:15:04,693][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:15:05,190][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:15:05,689][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:15:06,186][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:15:06,684][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:15:07,182][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:15:07,680][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:15:08,178][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:15:08,675][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:15:09,173][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:15:09,671][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21718 tokens. [2026-03-26 02:15:10,292][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 02:15:11,026][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:15:11,028][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:15:11,030][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:15:11,706][__main__][INFO] - Iteration 266 took 1m 13s (8.71% Gen, 90.37% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 45m 28s. Estimated total time: 61h 25m 58s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 51s, 500 more iterations: 10h 14m 19s. [2026-03-26 02:15:11,708][__main__][INFO] - Starting iteration 266. [2026-03-26 02:15:12,106][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:15:12,107][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:15:18,173][__main__][INFO] - Number of regex retries in iteration 266: 0 [2026-03-26 02:15:18,174][__main__][INFO] - agents played in iteration 266 are Alice, Bob [2026-03-26 02:15:19,048][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:15:19,093][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:15:19,094][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:15:19,094][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:15:19,971][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:15:20,427][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:15:20,931][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:15:21,429][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:15:21,928][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:15:22,426][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:15:22,924][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:15:23,423][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:15:23,923][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:15:24,424][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:15:24,924][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:15:25,424][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:15:25,924][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:15:26,423][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:15:26,923][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:15:27,423][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:15:27,923][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:15:28,421][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:15:28,919][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:15:29,418][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:15:29,917][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:15:30,414][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:15:30,914][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:15:31,413][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:15:31,912][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:15:32,411][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:15:32,910][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:15:33,409][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:15:33,908][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:15:34,407][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:15:34,905][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:15:35,405][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:15:35,904][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:15:36,404][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:15:36,903][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:15:37,402][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:15:37,900][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:15:38,399][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:15:38,898][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:15:39,397][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:15:39,895][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:15:40,394][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:15:40,891][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:15:41,389][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:15:41,886][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:15:42,382][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:15:42,881][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:15:43,379][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:15:43,877][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:15:44,375][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:15:44,874][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:15:45,372][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:15:45,870][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:15:46,368][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:15:46,866][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:15:47,365][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:15:47,864][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:15:48,365][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:15:48,865][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:15:49,364][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:15:49,863][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:15:50,362][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:15:50,863][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:15:51,363][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:15:51,863][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:15:52,362][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:15:52,861][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:15:53,361][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:15:53,860][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:15:54,359][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:15:54,858][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:15:55,358][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:15:55,857][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:15:56,356][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:15:56,856][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:15:57,356][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:15:57,856][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:15:58,357][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:15:58,857][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:15:59,358][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:15:59,858][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:16:00,356][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:16:00,854][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:16:01,352][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:16:01,850][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:16:02,348][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:16:02,846][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:16:03,344][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:16:03,842][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:16:04,341][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:16:04,841][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:16:05,341][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:16:05,841][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:16:06,340][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:16:06,840][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:16:07,341][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:16:07,841][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:16:08,340][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:16:08,839][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:16:09,338][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:16:09,838][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:16:10,336][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:16:10,835][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:16:11,334][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:16:11,833][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:16:12,330][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:16:12,828][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:16:13,326][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:16:13,824][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:16:14,322][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:16:14,820][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:16:15,319][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:16:15,820][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:16:16,320][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:16:16,819][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:16:17,317][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:16:17,818][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:16:18,318][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:16:18,819][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:16:19,317][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:16:19,816][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:16:20,316][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:16:20,816][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:16:21,315][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:16:21,816][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:16:22,315][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:16:22,814][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:16:23,313][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:16:23,813][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 02:16:24,434][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:16:25,280][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:16:25,282][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:16:25,284][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:16:25,953][__main__][INFO] - Iteration 267 took 1m 13s (8.22% Gen, 90.88% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 50m 40s. Estimated total time: 61h 32m 24s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 24s. [2026-03-26 02:16:25,956][__main__][INFO] - Starting iteration 267. [2026-03-26 02:16:26,354][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:16:26,355][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:16:32,815][__main__][INFO] - Number of regex retries in iteration 267: 0 [2026-03-26 02:16:32,816][__main__][INFO] - agents played in iteration 267 are Alice, Bob [2026-03-26 02:16:33,702][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:16:33,746][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:16:33,747][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:16:33,747][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:16:34,346][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:16:34,799][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:16:35,302][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:16:35,801][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:16:36,301][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:16:36,801][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:16:37,300][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:16:37,799][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:16:38,298][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:16:38,800][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:16:39,301][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:16:39,802][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:16:40,302][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:16:40,802][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:16:41,301][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:16:41,801][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:16:42,301][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:16:42,801][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:16:43,301][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:16:43,801][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:16:44,300][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:16:44,799][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:16:45,298][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:16:45,798][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:16:46,297][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:16:46,798][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:16:47,298][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:16:47,797][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:16:48,298][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:16:48,798][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:16:49,298][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:16:49,798][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:16:50,297][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:16:50,797][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:16:51,296][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:16:51,795][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:16:52,295][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:16:52,795][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:16:53,295][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:16:53,795][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:16:54,295][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:16:54,794][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:16:55,294][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:16:55,792][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:16:56,290][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:16:56,787][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:16:57,284][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:16:57,783][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:16:58,282][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:16:58,779][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:16:59,278][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:16:59,776][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:17:00,274][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:17:00,771][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:17:01,268][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:17:01,765][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:17:02,263][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:17:02,763][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:17:03,264][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:17:03,763][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:17:04,263][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:17:04,763][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:17:05,262][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:17:05,763][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:17:06,263][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:17:06,763][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:17:07,262][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:17:07,760][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:17:08,258][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:17:08,757][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:17:09,257][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:17:09,756][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:17:10,256][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:17:10,756][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:17:11,256][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:17:11,756][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:17:12,257][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:17:12,757][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:17:13,258][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:17:13,758][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:17:14,258][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:17:14,758][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:17:15,257][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:17:15,757][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:17:16,256][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:17:16,756][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:17:17,256][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:17:17,755][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:17:18,256][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:17:18,756][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:17:19,256][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:17:19,755][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:17:20,256][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:17:20,756][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:17:21,255][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:17:21,755][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:17:22,254][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:17:22,753][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:17:23,252][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:17:23,752][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:17:24,252][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:17:24,752][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:17:25,252][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:17:25,752][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:17:26,251][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:17:26,749][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:17:27,246][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:17:27,744][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:17:28,243][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:17:28,741][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:17:29,240][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:17:29,739][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:17:30,237][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:17:30,734][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:17:31,232][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:17:31,729][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:17:32,226][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:17:32,724][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:17:33,222][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:17:33,720][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:17:34,218][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:17:34,718][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:17:35,219][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:17:35,720][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:17:36,220][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:17:36,721][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:17:37,222][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:17:37,722][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:17:38,222][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:17:38,847][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:17:39,596][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:17:39,599][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:17:39,600][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:17:40,438][__main__][INFO] - Iteration 268 took 1m 14s (8.72% Gen, 90.15% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 1m 15s. Estimated total time: 61h 44m 13s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 28s, 500 more iterations: 10h 17m 22s. [2026-03-26 02:17:40,440][__main__][INFO] - Starting iteration 268. [2026-03-26 02:17:40,839][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:17:40,840][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:17:44,556][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:17:47,270][__main__][INFO] - Number of regex retries in iteration 268: 1 [2026-03-26 02:17:47,271][__main__][INFO] - agents played in iteration 268 are Alice, Bob [2026-03-26 02:17:48,188][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:17:48,232][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:17:48,233][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:17:48,233][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:17:48,842][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:17:49,296][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:17:49,800][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:17:50,298][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:17:50,796][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:17:51,295][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:17:51,793][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:17:52,291][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:17:52,789][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:17:53,289][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:17:53,787][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:17:54,285][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:17:54,784][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:17:55,283][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:17:55,782][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:17:56,280][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:17:56,779][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:17:57,278][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:17:57,777][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:17:58,276][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:17:58,775][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:17:59,274][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:17:59,772][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:18:00,271][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:18:00,770][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:18:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:18:01,768][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:18:02,267][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:18:02,766][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:18:03,265][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:18:03,764][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:18:04,263][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:18:04,762][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:18:05,261][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:18:05,759][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:18:06,258][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:18:06,756][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:18:07,255][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:18:07,753][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:18:08,251][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:18:08,749][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:18:09,249][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:18:09,749][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:18:10,249][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:18:10,748][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:18:11,248][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:18:11,747][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:18:12,246][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:18:12,745][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:18:13,245][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:18:13,744][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:18:14,245][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:18:14,744][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:18:15,244][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:18:15,743][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:18:16,243][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:18:16,743][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:18:17,242][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:18:17,740][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:18:18,242][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:18:18,741][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:18:19,240][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:18:19,738][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:18:20,237][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:18:20,737][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:18:21,235][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:18:21,734][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:18:22,233][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:18:22,731][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:18:23,230][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:18:23,727][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:18:24,226][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:18:24,725][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:18:25,223][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:18:25,722][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:18:26,221][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:18:26,720][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:18:27,219][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:18:27,718][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:18:28,217][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:18:28,717][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:18:29,216][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:18:29,716][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:18:30,216][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:18:30,714][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:18:31,214][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:18:31,713][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:18:32,212][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:18:32,711][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:18:33,209][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:18:33,708][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:18:34,207][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:18:34,706][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:18:35,204][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:18:35,703][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:18:36,202][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:18:36,701][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:18:37,201][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:18:37,699][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:18:38,197][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:18:38,696][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:18:39,194][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:18:39,691][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:18:40,190][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:18:40,688][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:18:41,187][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:18:41,686][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:18:42,185][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:18:42,683][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:18:43,182][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:18:43,681][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:18:44,181][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:18:44,681][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:18:45,179][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:18:45,678][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:18:46,176][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:18:46,675][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:18:47,173][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:18:47,672][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:18:48,173][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:18:48,672][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:18:49,171][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:18:49,669][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:18:50,167][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:18:50,666][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:18:51,164][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:18:51,662][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:18:52,160][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:18:52,659][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:18:53,296][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:18:54,045][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:18:54,047][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:18:54,049][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:18:54,709][__main__][INFO] - Iteration 269 took 1m 13s (8.71% Gen, 90.40% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 49m 18s. Estimated total time: 61h 33m 31s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 7s, 500 more iterations: 10h 15m 35s. [2026-03-26 02:18:54,711][__main__][INFO] - Starting iteration 269. [2026-03-26 02:18:55,116][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:18:55,117][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:18:56,167][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:19:01,390][__main__][INFO] - Number of regex retries in iteration 269: 1 [2026-03-26 02:19:01,391][__main__][INFO] - agents played in iteration 269 are Alice, Bob [2026-03-26 02:19:02,276][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:19:02,320][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:19:02,321][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:19:02,321][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:19:02,932][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:19:03,385][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:19:03,888][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:19:04,386][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:19:04,883][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:19:05,381][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:19:05,879][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:19:06,376][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:19:06,875][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:19:07,375][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:19:07,875][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:19:08,373][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:19:08,870][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:19:09,368][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:19:09,866][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:19:10,363][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:19:10,862][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:19:11,362][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:19:11,862][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:19:12,361][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:19:12,859][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:19:13,359][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:19:13,858][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:19:14,357][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:19:14,857][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:19:15,377][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:19:15,884][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:19:16,385][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:19:16,885][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:19:17,385][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:19:17,886][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:19:18,386][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:19:18,888][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:19:19,387][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:19:19,885][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:19:20,384][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:19:20,883][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:19:21,382][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:19:21,881][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:19:22,379][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:19:22,877][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:19:23,381][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:19:23,884][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:19:24,385][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:19:24,886][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:19:25,388][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:19:25,888][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:19:26,388][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:19:26,890][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:19:27,389][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:19:27,887][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:19:28,385][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:19:28,883][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:19:29,382][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:19:29,880][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:19:30,378][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:19:30,878][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:19:31,377][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:19:31,876][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:19:32,375][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:19:32,873][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:19:33,372][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:19:33,870][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:19:34,368][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:19:34,867][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:19:35,364][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:19:35,861][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:19:36,358][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:19:36,856][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:19:37,352][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:19:37,849][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:19:38,347][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:19:38,845][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:19:39,344][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:19:39,842][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:19:40,341][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:19:40,839][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:19:41,338][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:19:41,835][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:19:42,333][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:19:42,831][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:19:43,329][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:19:43,828][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:19:44,327][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:19:44,827][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:19:45,325][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:19:45,824][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:19:46,323][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:19:46,822][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:19:47,322][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:19:47,822][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:19:48,325][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:19:48,825][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:19:49,325][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:19:49,825][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:19:50,324][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:19:50,824][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:19:51,323][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:19:51,822][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:19:52,322][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:19:52,820][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:19:53,320][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:19:53,820][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:19:54,318][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:19:54,818][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:19:55,321][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:19:55,822][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:19:56,324][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:19:56,825][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:19:57,326][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:19:57,828][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:19:58,329][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:19:58,831][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:19:59,329][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:19:59,826][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:20:00,324][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:20:00,822][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:20:01,322][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:20:01,820][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:20:02,319][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:20:02,817][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:20:03,319][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:20:03,818][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:20:04,317][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:20:04,819][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:20:05,319][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:20:05,818][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:20:06,318][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:20:06,817][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 02:20:07,448][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 02:20:08,185][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:20:08,188][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:20:08,189][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:20:08,859][__main__][INFO] - Iteration 270 took 1m 13s (8.51% Gen, 90.58% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 41m 42s. Estimated total time: 61h 27m 9s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 54s, 500 more iterations: 10h 14m 31s. [2026-03-26 02:20:08,861][__main__][INFO] - Starting iteration 270. [2026-03-26 02:20:09,260][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:20:09,260][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:20:15,499][__main__][INFO] - Number of regex retries in iteration 270: 0 [2026-03-26 02:20:15,500][__main__][INFO] - agents played in iteration 270 are Alice, Bob [2026-03-26 02:20:16,389][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:20:16,710][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:20:16,711][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:20:16,712][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:20:17,309][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:20:17,763][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:20:18,268][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:20:18,769][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:20:19,268][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:20:19,766][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:20:20,264][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:20:20,763][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:20:21,262][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:20:21,760][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:20:22,258][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:20:22,757][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:20:23,255][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:20:23,754][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:20:24,252][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:20:24,750][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:20:25,249][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:20:25,749][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:20:26,248][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:20:26,747][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:20:27,247][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:20:27,747][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:20:28,246][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:20:28,745][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:20:29,244][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:20:29,743][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:20:30,242][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:20:30,742][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:20:31,241][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:20:31,740][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:20:32,239][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:20:32,737][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:20:33,236][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:20:33,735][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:20:34,234][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:20:34,732][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:20:35,230][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:20:35,728][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:20:36,225][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:20:36,723][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:20:37,222][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:20:37,720][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:20:38,219][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:20:38,717][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:20:39,217][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:20:39,716][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:20:40,215][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:20:40,713][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:20:41,210][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:20:41,708][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:20:42,207][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:20:42,705][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:20:43,205][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:20:43,703][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:20:44,202][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:20:44,701][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:20:45,200][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:20:45,704][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:20:46,206][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:20:46,706][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:20:47,205][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:20:47,705][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:20:48,207][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:20:48,708][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:20:49,207][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:20:49,706][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:20:50,205][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:20:50,704][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:20:51,203][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:20:51,702][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:20:52,201][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:20:52,700][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:20:53,199][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:20:53,697][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:20:54,195][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:20:54,694][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:20:55,192][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:20:55,689][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:20:56,187][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:20:56,684][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:20:57,182][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:20:57,681][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:20:58,180][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:20:58,679][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:20:59,178][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:20:59,678][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:21:00,176][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:21:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:21:01,175][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:21:01,675][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:21:02,174][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:21:02,672][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:21:03,171][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:21:03,668][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:21:04,167][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:21:04,667][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:21:05,166][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:21:05,663][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:21:06,163][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:21:06,661][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:21:07,159][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:21:07,655][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:21:08,152][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:21:08,649][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:21:09,147][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:21:09,644][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:21:10,143][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:21:10,642][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:21:11,141][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:21:11,639][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:21:12,136][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:21:12,634][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:21:13,132][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:21:13,631][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:21:14,130][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:21:14,628][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:21:15,127][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:21:15,625][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:21:16,124][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:21:16,622][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:21:17,121][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:21:17,620][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:21:18,121][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:21:18,622][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:21:19,122][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:21:19,622][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:21:20,123][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:21:20,622][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:21:21,123][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 02:21:21,763][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:21:22,507][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:21:22,509][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:21:22,510][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:21:23,161][__main__][INFO] - Iteration 271 took 1m 13s (8.44% Gen, 90.67% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 48m 26s. Estimated total time: 61h 35m 7s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 10s, 500 more iterations: 10h 15m 51s. [2026-03-26 02:21:23,164][__main__][INFO] - Starting iteration 271. [2026-03-26 02:21:23,564][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:21:23,564][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:21:29,973][__main__][INFO] - Number of regex retries in iteration 271: 0 [2026-03-26 02:21:29,974][__main__][INFO] - agents played in iteration 271 are Alice, Bob [2026-03-26 02:21:30,872][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:21:30,917][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:21:30,918][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:21:30,918][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:21:31,519][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:21:31,979][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:21:32,484][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:21:32,983][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:21:33,482][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:21:33,982][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:21:34,481][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:21:34,980][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:21:35,478][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:21:35,977][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:21:36,477][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:21:36,977][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:21:37,476][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:21:37,974][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:21:38,472][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:21:38,971][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:21:39,470][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:21:39,968][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:21:40,467][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:21:40,965][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:21:41,463][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:21:41,962][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:21:42,461][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:21:42,960][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:21:43,459][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:21:43,959][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:21:44,458][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:21:44,956][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:21:45,455][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:21:45,953][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:21:46,451][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:21:46,949][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:21:47,448][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:21:47,946][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:21:48,448][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:21:48,946][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:21:49,443][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:21:49,941][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:21:50,440][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:21:50,938][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:21:51,438][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:21:51,936][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:21:52,435][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:21:52,934][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:21:53,434][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:21:53,932][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:21:54,430][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:21:54,927][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:21:55,426][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:21:55,926][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:21:56,426][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:21:56,925][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:21:57,424][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:21:57,923][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:21:58,422][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:21:58,921][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:21:59,420][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:21:59,919][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:22:00,417][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:22:00,917][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:22:01,415][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:22:01,913][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:22:02,411][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:22:02,909][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:22:03,408][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:22:03,905][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:22:04,404][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:22:04,903][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:22:05,401][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:22:05,901][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:22:06,401][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:22:06,900][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:22:07,399][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:22:07,898][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:22:08,397][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:22:08,897][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:22:09,396][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:22:09,894][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:22:10,393][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:22:10,892][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:22:11,391][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:22:11,889][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:22:12,388][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:22:12,886][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:22:13,383][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:22:13,881][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:22:14,379][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:22:14,877][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:22:15,379][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:22:15,880][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:22:16,379][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:22:16,878][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:22:17,376][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:22:17,875][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:22:18,374][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:22:18,872][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:22:19,371][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:22:19,868][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:22:20,366][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:22:20,863][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:22:21,361][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:22:21,859][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:22:22,356][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:22:22,853][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:22:23,350][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:22:23,848][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:22:24,345][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:22:24,842][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:22:25,339][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:22:25,837][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:22:26,334][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:22:26,831][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:22:27,328][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:22:27,826][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:22:28,325][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:22:28,824][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:22:29,324][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:22:29,822][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:22:30,321][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:22:30,819][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:22:31,319][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:22:31,817][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:22:32,318][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:22:32,816][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:22:33,315][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:22:33,813][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:22:34,311][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:22:34,808][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:22:35,306][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 02:22:35,931][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 02:22:36,664][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:22:36,666][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:22:36,668][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:22:37,315][__main__][INFO] - Iteration 272 took 1m 13s (8.69% Gen, 90.43% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 39m 40s. Estimated total time: 61h 27m 36s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 55s, 500 more iterations: 10h 14m 36s. [2026-03-26 02:22:37,317][__main__][INFO] - Starting iteration 272. [2026-03-26 02:22:37,715][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:22:37,716][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:22:38,296][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:22:43,939][__main__][INFO] - Number of regex retries in iteration 272: 1 [2026-03-26 02:22:43,940][__main__][INFO] - agents played in iteration 272 are Alice, Bob [2026-03-26 02:22:44,812][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:22:44,857][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:22:44,858][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:22:44,858][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:22:45,465][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:22:45,916][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:22:46,419][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:22:46,915][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:22:47,412][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:22:47,908][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:22:48,406][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:22:48,903][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:22:49,400][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:22:49,901][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:22:50,400][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:22:50,898][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:22:51,396][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:22:51,895][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:22:52,393][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:22:52,892][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:22:53,391][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:22:53,888][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:22:54,386][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:22:54,884][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:22:55,382][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:22:55,880][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:22:56,378][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:22:56,875][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:22:57,373][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:22:57,872][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:22:58,371][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:22:58,870][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:22:59,368][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:22:59,867][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:23:00,365][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:23:00,863][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:23:01,363][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:23:01,881][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:23:02,384][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:23:02,884][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:23:03,383][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:23:03,882][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:23:04,381][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:23:04,880][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:23:05,379][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:23:05,879][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:23:06,379][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:23:06,879][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:23:07,378][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:23:07,878][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:23:08,377][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:23:08,876][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:23:09,375][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:23:09,874][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:23:10,374][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:23:10,873][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:23:11,372][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:23:11,872][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:23:12,372][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:23:12,871][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:23:13,371][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:23:13,870][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:23:14,368][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:23:14,866][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:23:15,365][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:23:15,863][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:23:16,361][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:23:16,860][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:23:17,358][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:23:17,854][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:23:18,352][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:23:18,849][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:23:19,345][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:23:19,842][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:23:20,339][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:23:20,836][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:23:21,332][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:23:21,832][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:23:22,330][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:23:22,828][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:23:23,326][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:23:23,825][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:23:24,323][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:23:24,822][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:23:25,321][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:23:25,820][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:23:26,318][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:23:26,815][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:23:27,313][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:23:27,811][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:23:28,308][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:23:28,806][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:23:29,304][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:23:29,803][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:23:30,302][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:23:30,800][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:23:31,299][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:23:31,798][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:23:32,297][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:23:32,798][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:23:33,297][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:23:33,796][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:23:34,296][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:23:34,794][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:23:35,292][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:23:35,791][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:23:36,289][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:23:36,788][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:23:37,286][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:23:37,785][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:23:38,283][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:23:38,783][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:23:39,282][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:23:39,781][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:23:40,280][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:23:40,779][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:23:41,278][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:23:41,778][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:23:42,277][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:23:42,777][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:23:43,276][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:23:43,775][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:23:44,277][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:23:44,777][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:23:45,277][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:23:45,776][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:23:46,274][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:23:46,772][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:23:47,270][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:23:47,768][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:23:48,269][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:23:48,767][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:23:49,266][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 02:23:49,896][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:23:50,632][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:23:50,634][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:23:50,636][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:23:51,464][__main__][INFO] - Iteration 273 took 1m 13s (8.44% Gen, 90.44% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 38m 19s. Estimated total time: 61h 27m 29s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 54s, 500 more iterations: 10h 14m 34s. [2026-03-26 02:23:51,467][__main__][INFO] - Starting iteration 273. [2026-03-26 02:23:51,865][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:23:51,866][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:23:57,882][__main__][INFO] - Number of regex retries in iteration 273: 0 [2026-03-26 02:23:57,883][__main__][INFO] - agents played in iteration 273 are Alice, Bob [2026-03-26 02:23:58,768][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:23:58,813][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:23:58,813][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:23:58,814][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:23:59,423][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:23:59,879][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:24:00,383][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:24:00,884][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:24:01,384][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:24:01,884][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:24:02,384][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:24:02,884][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:24:03,385][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:24:03,884][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:24:04,382][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:24:04,881][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:24:05,378][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:24:05,875][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:24:06,373][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:24:06,871][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:24:07,368][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:24:07,868][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:24:08,367][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:24:08,867][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:24:09,366][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:24:09,865][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:24:10,364][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:24:10,863][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:24:11,363][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:24:11,862][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:24:12,361][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:24:12,860][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:24:13,358][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:24:13,856][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:24:14,354][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:24:14,851][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:24:15,349][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:24:15,848][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:24:16,346][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:24:16,844][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:24:17,343][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:24:17,842][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:24:18,343][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:24:18,843][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:24:19,342][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:24:19,842][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:24:20,341][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:24:20,841][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:24:21,340][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:24:21,839][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:24:22,337][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:24:22,836][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:24:23,336][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:24:23,835][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:24:24,333][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:24:24,832][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:24:25,332][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:24:25,829][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:24:26,327][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:24:26,825][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:24:27,325][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:24:27,825][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:24:28,325][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:24:28,826][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:24:29,325][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:24:29,825][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:24:30,325][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:24:30,825][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:24:31,325][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:24:31,827][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:24:32,327][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:24:32,827][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:24:33,326][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:24:33,827][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:24:34,327][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:24:34,828][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:24:35,328][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:24:35,826][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:24:36,324][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:24:36,822][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:24:37,321][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:24:37,819][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:24:38,319][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:24:38,818][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:24:39,316][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:24:39,815][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:24:40,313][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:24:40,811][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:24:41,308][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:24:41,806][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:24:42,304][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:24:42,802][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:24:43,301][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:24:43,799][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:24:44,296][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:24:44,793][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:24:45,291][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:24:45,788][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:24:46,285][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:24:46,783][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:24:47,281][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:24:47,779][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:24:48,279][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:24:48,777][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:24:49,275][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:24:49,773][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:24:50,270][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:24:50,768][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:24:51,267][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:24:51,765][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:24:52,265][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:24:52,764][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:24:53,263][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:24:53,761][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:24:54,261][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:24:54,760][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:24:55,260][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:24:55,759][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:24:56,257][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:24:56,755][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:24:57,253][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:24:57,751][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:24:58,249][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:24:58,747][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:24:59,246][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:24:59,745][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:25:00,244][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:25:00,744][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:25:01,244][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:25:01,743][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:25:02,242][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:25:02,741][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:25:03,240][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21760 tokens. [2026-03-26 02:25:03,873][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 02:25:04,606][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:25:04,608][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:25:04,610][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:25:05,257][__main__][INFO] - Iteration 274 took 1m 13s (8.20% Gen, 90.92% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 19m 14s. Estimated total time: 61h 9m 37s. Time estimates for 10 more iterations: 12m 13s, 100 more iterations: 2h 2m 19s, 500 more iterations: 10h 11m 36s. [2026-03-26 02:25:05,259][__main__][INFO] - Starting iteration 274. [2026-03-26 02:25:05,657][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:25:05,658][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:25:11,879][__main__][INFO] - Number of regex retries in iteration 274: 0 [2026-03-26 02:25:11,880][__main__][INFO] - agents played in iteration 274 are Alice, Bob [2026-03-26 02:25:12,785][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:25:12,830][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:25:12,831][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:25:12,831][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:25:13,441][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:25:13,895][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:25:14,398][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:25:14,895][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:25:15,392][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:25:15,889][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:25:16,386][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:25:16,884][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:25:17,381][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:25:17,887][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:25:18,389][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:25:18,888][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:25:19,386][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:25:19,886][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:25:20,384][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:25:20,884][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:25:21,384][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:25:21,884][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:25:22,384][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:25:22,884][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:25:23,383][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:25:23,883][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:25:24,384][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:25:24,884][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:25:25,384][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:25:25,884][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:25:26,383][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:25:26,884][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:25:27,383][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:25:27,883][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:25:28,382][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:25:28,882][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:25:29,381][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:25:29,880][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:25:30,378][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:25:30,877][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:25:31,377][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:25:31,877][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:25:32,376][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:25:32,876][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:25:33,375][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:25:33,892][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:25:34,397][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:25:34,898][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:25:35,398][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:25:35,898][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:25:36,399][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:25:36,899][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:25:37,401][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:25:37,901][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:25:38,401][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:25:38,901][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:25:39,401][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:25:39,901][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:25:40,400][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:25:40,901][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:25:41,401][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:25:41,899][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:25:42,397][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:25:42,895][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:25:43,392][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:25:43,889][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:25:44,386][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:25:44,884][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:25:45,382][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:25:45,881][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:25:46,379][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:25:46,878][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:25:47,377][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:25:47,876][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:25:48,376][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:25:48,875][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:25:49,373][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:25:49,873][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:25:50,373][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:25:50,872][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:25:51,371][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:25:51,870][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:25:52,369][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:25:52,868][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:25:53,368][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:25:53,868][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:25:54,368][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:25:54,867][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:25:55,367][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:25:55,867][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:25:56,367][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:25:56,868][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:25:57,368][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:25:57,867][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:25:58,367][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:25:58,867][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:25:59,366][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:25:59,865][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:26:00,365][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:26:00,864][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:26:01,364][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:26:01,863][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:26:02,361][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:26:02,860][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:26:03,360][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:26:03,859][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:26:04,359][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:26:04,859][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:26:05,359][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:26:05,860][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:26:06,361][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:26:06,862][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:26:07,363][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:26:07,864][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:26:08,364][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:26:08,865][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:26:09,364][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:26:09,864][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:26:10,365][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:26:10,865][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:26:11,365][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:26:11,864][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:26:12,366][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:26:12,864][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:26:13,364][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:26:13,862][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:26:14,360][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:26:14,859][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:26:15,357][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:26:15,855][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:26:16,353][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:26:16,851][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:26:17,349][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:26:17,977][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:26:18,715][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:26:18,717][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:26:18,719][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:26:19,385][__main__][INFO] - Iteration 275 took 1m 13s (8.44% Gen, 90.66% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 34m 47s. Estimated total time: 61h 26m 24s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 52s, 500 more iterations: 10h 14m 24s. [2026-03-26 02:26:19,387][__main__][INFO] - Starting iteration 275. [2026-03-26 02:26:19,785][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:26:19,786][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:26:26,129][__main__][INFO] - Number of regex retries in iteration 275: 0 [2026-03-26 02:26:26,130][__main__][INFO] - agents played in iteration 275 are Alice, Bob [2026-03-26 02:26:27,005][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:26:27,049][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:26:27,050][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:26:27,050][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:26:27,654][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:26:28,108][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:26:28,611][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:26:29,110][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:26:29,609][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:26:30,108][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:26:30,605][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:26:31,108][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:26:31,607][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:26:32,108][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:26:32,607][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:26:33,108][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:26:33,608][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:26:34,107][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:26:34,608][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:26:35,108][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:26:35,608][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:26:36,106][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:26:36,605][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:26:37,105][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:26:37,605][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:26:38,105][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:26:38,603][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:26:39,102][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:26:39,601][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:26:40,102][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:26:40,601][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:26:41,102][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:26:41,602][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:26:42,102][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:26:42,601][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:26:43,102][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:26:43,602][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:26:44,101][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:26:44,601][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:26:45,100][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:26:45,602][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:26:46,102][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:26:46,601][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:26:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:26:47,600][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:26:48,100][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:26:48,602][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:26:49,102][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:26:49,601][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:26:50,101][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:26:50,601][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:26:51,101][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:26:51,602][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:26:52,101][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:26:52,600][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:26:53,098][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:26:53,597][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:26:54,095][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:26:54,594][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:26:55,094][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:26:55,593][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:26:56,093][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:26:56,593][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:26:57,092][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:26:57,591][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:26:58,091][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:26:58,589][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:26:59,089][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:26:59,588][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:27:00,087][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:27:00,585][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:27:01,083][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:27:01,582][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:27:02,081][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:27:02,579][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:27:03,078][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:27:03,577][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:27:04,077][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:27:04,578][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:27:05,077][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:27:05,576][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:27:06,076][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:27:06,575][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:27:07,076][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:27:07,575][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:27:08,074][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:27:08,572][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:27:09,070][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:27:09,570][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:27:10,069][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:27:10,567][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:27:11,067][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:27:11,566][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:27:12,066][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:27:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:27:13,066][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:27:13,568][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:27:14,068][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:27:14,568][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:27:15,068][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:27:15,568][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:27:16,066][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:27:16,565][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:27:17,063][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:27:17,562][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:27:18,062][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:27:18,563][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:27:19,062][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:27:19,560][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:27:20,059][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:27:20,558][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:27:21,056][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:27:21,556][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:27:22,057][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:27:22,556][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:27:23,056][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:27:23,555][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:27:24,053][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:27:24,552][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:27:25,050][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:27:25,549][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:27:26,048][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:27:26,547][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:27:27,045][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:27:27,545][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:27:28,045][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:27:28,545][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:27:29,045][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:27:29,545][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:27:30,046][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:27:30,546][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:27:31,045][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:27:31,545][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21739 tokens. [2026-03-26 02:27:32,173][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:27:32,930][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:27:32,933][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:27:32,934][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:27:33,587][__main__][INFO] - Iteration 276 took 1m 13s (8.60% Gen, 90.52% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 37m 16s. Estimated total time: 61h 30m 7s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 0s, 500 more iterations: 10h 15m 1s. [2026-03-26 02:27:33,590][__main__][INFO] - Starting iteration 276. [2026-03-26 02:27:33,988][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:27:33,989][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:27:40,454][__main__][INFO] - Number of regex retries in iteration 276: 0 [2026-03-26 02:27:40,455][__main__][INFO] - agents played in iteration 276 are Alice, Bob [2026-03-26 02:27:41,352][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:27:41,397][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:27:41,398][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:27:41,398][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:27:42,001][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:27:42,456][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:27:42,959][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:27:43,458][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:27:43,960][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:27:44,459][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:27:44,959][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:27:45,458][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:27:45,957][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:27:46,458][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:27:46,957][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:27:47,458][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:27:47,959][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:27:48,462][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:27:48,961][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:27:49,461][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:27:49,961][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:27:50,460][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:27:50,959][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:27:51,458][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:27:51,957][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:27:52,457][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:27:52,955][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:27:53,453][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:27:53,951][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:27:54,450][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:27:54,948][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:27:55,447][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:27:55,945][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:27:56,444][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:27:56,943][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:27:57,441][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:27:57,940][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:27:58,438][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:27:58,938][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:27:59,437][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:27:59,937][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:28:00,435][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:28:00,933][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:28:01,432][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:28:01,930][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:28:02,428][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:28:02,925][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:28:03,425][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:28:03,923][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:28:04,422][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:28:04,920][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:28:05,418][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:28:05,916][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:28:06,415][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:28:06,914][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:28:07,413][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:28:07,912][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:28:08,411][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:28:08,910][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:28:09,409][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:28:09,908][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:28:10,409][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:28:10,909][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:28:11,408][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:28:11,909][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:28:12,408][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:28:12,908][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:28:13,409][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:28:13,909][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:28:14,408][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:28:14,909][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:28:15,410][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:28:15,911][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:28:16,410][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:28:16,909][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:28:17,408][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:28:17,908][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:28:18,410][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:28:18,910][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:28:19,409][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:28:19,908][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:28:20,407][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:28:20,906][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:28:21,406][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:28:21,906][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:28:22,405][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:28:22,903][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:28:23,401][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:28:23,900][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:28:24,399][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:28:24,898][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:28:25,397][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:28:25,895][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:28:26,393][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:28:26,892][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:28:27,390][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:28:27,889][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:28:28,387][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:28:28,886][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:28:29,385][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:28:29,884][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:28:30,382][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:28:30,881][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:28:31,380][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:28:31,878][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:28:32,377][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:28:32,875][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:28:33,373][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:28:33,871][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:28:34,369][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:28:34,868][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:28:35,366][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:28:35,863][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:28:36,360][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:28:36,859][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:28:37,356][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:28:37,854][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:28:38,353][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:28:38,852][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:28:39,352][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:28:39,851][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:28:40,350][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:28:40,848][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:28:41,347][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:28:41,846][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:28:42,346][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:28:42,846][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:28:43,347][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:28:43,846][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:28:44,346][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:28:44,846][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:28:45,346][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:28:45,846][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 02:28:46,480][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:28:47,225][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:28:47,228][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:28:47,230][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:28:47,883][__main__][INFO] - Iteration 277 took 1m 13s (8.75% Gen, 90.36% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 40m 39s. Estimated total time: 61h 34m 45s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 9s, 500 more iterations: 10h 15m 47s. [2026-03-26 02:28:47,885][__main__][INFO] - Starting iteration 277. [2026-03-26 02:28:48,296][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:28:48,297][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:28:49,375][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 0 books, 30 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:28:51,057][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:28:54,839][__main__][INFO] - Number of regex retries in iteration 277: 2 [2026-03-26 02:28:54,840][__main__][INFO] - agents played in iteration 277 are Alice, Bob [2026-03-26 02:28:55,736][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:28:55,780][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:28:55,781][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:28:55,781][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:28:56,383][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:28:56,839][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:28:57,344][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:28:57,844][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:28:58,343][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:28:58,843][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:28:59,343][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:28:59,843][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:29:00,342][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:29:00,842][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:29:01,341][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:29:01,839][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:29:02,338][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:29:02,836][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:29:03,336][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:29:03,835][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:29:04,334][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:29:04,833][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:29:05,331][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:29:05,829][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:29:06,327][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:29:06,825][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:29:07,322][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:29:07,822][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:29:08,320][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:29:08,820][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:29:09,319][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:29:09,817][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:29:10,315][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:29:10,812][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:29:11,309][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:29:11,807][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:29:12,306][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:29:12,803][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:29:13,300][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:29:13,797][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:29:14,294][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:29:14,793][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:29:15,290][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:29:15,789][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:29:16,288][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:29:16,786][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:29:17,284][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:29:17,782][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:29:18,280][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:29:18,780][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:29:19,278][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:29:19,775][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:29:20,272][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:29:20,771][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:29:21,269][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:29:21,768][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:29:22,268][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:29:22,767][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:29:23,266][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:29:23,765][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:29:24,263][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:29:24,763][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:29:25,263][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:29:25,763][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:29:26,263][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:29:26,762][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:29:27,263][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:29:27,762][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:29:28,263][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:29:28,763][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:29:29,263][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:29:29,762][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:29:30,262][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:29:30,761][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:29:31,261][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:29:31,762][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:29:32,263][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:29:32,762][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:29:33,260][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:29:33,759][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:29:34,259][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:29:34,757][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:29:35,257][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:29:35,757][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:29:36,257][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:29:36,755][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:29:37,253][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:29:37,751][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:29:38,248][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:29:38,746][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:29:39,243][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:29:39,742][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:29:40,240][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:29:40,740][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:29:41,239][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:29:41,739][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:29:42,237][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:29:42,736][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:29:43,234][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:29:43,734][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:29:44,233][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:29:44,731][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:29:45,229][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:29:45,730][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:29:46,229][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:29:46,726][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:29:47,223][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:29:47,721][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:29:48,222][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:29:48,721][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:29:49,219][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:29:49,717][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:29:50,215][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:29:50,713][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:29:51,211][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:29:51,709][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:29:52,208][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:29:52,708][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:29:53,206][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:29:53,706][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:29:54,206][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:29:54,704][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:29:55,204][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:29:55,702][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:29:56,201][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:29:56,701][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:29:57,200][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:29:57,700][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:29:58,200][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:29:58,700][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:29:59,202][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:29:59,702][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:30:00,202][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 02:30:00,825][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 02:30:01,558][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:30:01,560][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:30:01,561][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:30:02,209][__main__][INFO] - Iteration 278 took 1m 13s (8.85% Gen, 90.27% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 40m 22s. Estimated total time: 61h 35m 42s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 11s, 500 more iterations: 10h 15m 57s. [2026-03-26 02:30:02,212][__main__][INFO] - Starting iteration 278. [2026-03-26 02:30:02,610][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:30:02,611][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:30:09,151][__main__][INFO] - Number of regex retries in iteration 278: 0 [2026-03-26 02:30:09,152][__main__][INFO] - agents played in iteration 278 are Alice, Bob [2026-03-26 02:30:10,024][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:30:10,068][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:30:10,069][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:30:10,069][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:30:10,675][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:30:11,132][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:30:11,635][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:30:12,134][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:30:12,632][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:30:13,132][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:30:13,631][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:30:14,130][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:30:14,628][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:30:15,127][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:30:15,627][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:30:16,126][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:30:16,625][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:30:17,123][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:30:17,622][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:30:18,121][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:30:18,622][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:30:19,121][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:30:19,622][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:30:20,121][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:30:20,620][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:30:21,119][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:30:21,618][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:30:22,117][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:30:22,615][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:30:23,115][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:30:23,614][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:30:24,113][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:30:24,612][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:30:25,112][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:30:25,612][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:30:26,111][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:30:26,611][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:30:27,110][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:30:27,609][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:30:28,109][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:30:28,608][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:30:29,107][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:30:29,607][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:30:30,106][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:30:30,606][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:30:31,104][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:30:31,603][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:30:32,102][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:30:32,601][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:30:33,099][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:30:33,598][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:30:34,095][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:30:34,594][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:30:35,089][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:30:35,584][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:30:36,079][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:30:36,574][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:30:37,069][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:30:37,564][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:30:38,060][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:30:38,555][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:30:39,053][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:30:39,553][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:30:40,052][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:30:40,552][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:30:41,050][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:30:41,549][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:30:42,049][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:30:42,549][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:30:43,048][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:30:43,547][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:30:44,045][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:30:44,545][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:30:45,043][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:30:45,542][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:30:46,041][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:30:46,540][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:30:47,038][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:30:47,537][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:30:48,036][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:30:48,535][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:30:49,034][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:30:49,532][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:30:50,029][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:30:50,528][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:30:51,026][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:30:51,526][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:30:52,024][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:30:52,523][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:30:53,021][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:30:53,522][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:30:54,020][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:30:54,519][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:30:55,019][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:30:55,518][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:30:56,017][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:30:56,516][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:30:57,015][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:30:57,514][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:30:58,013][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:30:58,513][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:30:59,012][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:30:59,512][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:31:00,011][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:31:00,511][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:31:01,010][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:31:01,510][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:31:02,010][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:31:02,509][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:31:03,008][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:31:03,507][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:31:04,006][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:31:04,504][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:31:05,004][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:31:05,502][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:31:06,001][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:31:06,500][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:31:06,995][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:31:07,490][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:31:07,986][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:31:08,481][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:31:08,975][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:31:09,469][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:31:09,964][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:31:10,459][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:31:10,959][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:31:11,457][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:31:11,956][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:31:12,454][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:31:12,953][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:31:13,451][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:31:13,949][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:31:14,448][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:31:15,079][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:31:15,864][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:31:15,869][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:31:15,870][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:31:16,894][__main__][INFO] - Iteration 279 took 1m 14s (8.81% Gen, 89.81% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 57m 38s. Estimated total time: 61h 54m 12s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 48s, 500 more iterations: 10h 19m 2s. [2026-03-26 02:31:16,897][__main__][INFO] - Starting iteration 279. [2026-03-26 02:31:17,296][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:31:17,297][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:31:23,710][__main__][INFO] - Number of regex retries in iteration 279: 0 [2026-03-26 02:31:23,710][__main__][INFO] - agents played in iteration 279 are Alice, Bob [2026-03-26 02:31:24,599][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:31:24,644][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:31:24,644][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:31:24,645][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:31:25,250][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:31:25,704][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:31:26,209][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:31:26,707][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:31:27,207][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:31:27,706][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:31:28,205][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:31:28,704][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:31:29,204][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:31:29,703][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:31:30,201][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:31:30,700][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:31:31,197][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:31:31,695][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:31:32,192][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:31:32,688][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:31:33,186][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:31:33,686][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:31:34,184][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:31:34,682][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:31:35,180][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:31:35,679][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:31:36,178][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:31:36,676][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:31:37,174][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:31:37,674][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:31:38,173][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:31:38,671][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:31:39,170][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:31:39,669][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:31:40,168][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:31:40,666][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:31:41,164][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:31:41,665][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:31:42,165][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:31:42,664][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:31:43,164][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:31:43,663][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:31:44,162][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:31:44,662][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:31:45,161][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:31:45,659][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:31:46,156][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:31:46,653][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:31:47,150][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:31:47,648][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:31:48,148][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:31:48,647][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:31:49,145][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:31:49,642][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:31:50,141][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:31:50,640][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:31:51,139][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:31:51,638][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:31:52,136][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:31:52,635][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:31:53,132][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:31:53,632][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:31:54,131][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:31:54,631][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:31:55,130][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:31:55,630][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:31:56,130][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:31:56,629][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:31:57,128][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:31:57,628][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:31:58,127][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:31:58,627][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:31:59,126][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:31:59,625][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:32:00,124][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:32:00,623][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:32:01,123][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:32:01,621][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:32:02,120][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:32:02,619][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:32:03,116][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:32:03,614][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:32:04,112][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:32:04,609][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:32:05,106][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:32:05,604][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:32:06,103][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:32:06,603][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:32:07,101][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:32:07,600][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:32:08,099][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:32:08,596][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:32:09,094][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:32:09,592][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:32:10,090][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:32:10,589][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:32:11,088][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:32:11,587][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:32:12,086][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:32:12,585][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:32:13,084][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:32:13,583][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:32:14,082][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:32:14,583][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:32:15,082][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:32:15,582][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:32:16,082][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:32:16,582][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:32:17,082][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:32:17,582][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:32:18,082][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:32:18,581][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:32:19,079][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:32:19,578][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:32:20,076][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:32:20,574][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:32:21,071][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:32:21,568][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:32:22,066][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:32:22,563][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:32:23,061][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:32:23,559][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:32:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:32:24,555][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:32:25,053][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:32:25,554][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:32:26,053][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:32:26,553][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:32:27,052][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:32:27,552][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:32:28,051][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:32:28,551][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:32:29,050][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:32:29,682][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:32:30,428][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:32:30,431][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:32:30,433][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:32:31,267][__main__][INFO] - Iteration 280 took 1m 13s (8.67% Gen, 90.20% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 40m 44s. Estimated total time: 61h 38m 33s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 17s, 500 more iterations: 10h 16m 25s. [2026-03-26 02:32:31,269][__main__][INFO] - Starting iteration 280. [2026-03-26 02:32:31,669][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:32:31,669][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:32:32,262][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:32:34,385][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:32:38,543][__main__][INFO] - Number of regex retries in iteration 280: 2 [2026-03-26 02:32:38,544][__main__][INFO] - agents played in iteration 280 are Alice, Bob [2026-03-26 02:32:39,473][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:32:39,517][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:32:39,518][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:32:39,518][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:32:40,122][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:32:40,578][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:32:41,082][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:32:41,581][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:32:42,079][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:32:42,578][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:32:43,078][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:32:43,576][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:32:44,075][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:32:44,573][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:32:45,072][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:32:45,573][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:32:46,073][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:32:46,572][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:32:47,070][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:32:47,568][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:32:48,067][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:32:48,570][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:32:49,068][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:32:49,568][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:32:50,067][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:32:50,567][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:32:51,066][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:32:51,565][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:32:52,064][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:32:52,565][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:32:53,067][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:32:53,567][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:32:54,067][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:32:54,568][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:32:55,069][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:32:55,570][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:32:56,070][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:32:56,568][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:32:57,068][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:32:57,567][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:32:58,066][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:32:58,564][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:32:59,063][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:32:59,561][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:33:00,060][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:33:00,559][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:33:01,058][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:33:01,558][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:33:02,059][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:33:02,558][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:33:03,059][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:33:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:33:04,057][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:33:04,558][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:33:05,058][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:33:05,559][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:33:06,059][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:33:06,559][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:33:07,059][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:33:07,559][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:33:08,060][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:33:08,558][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:33:09,057][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:33:09,555][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:33:10,053][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:33:10,550][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:33:11,047][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:33:11,547][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:33:12,045][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:33:12,545][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:33:13,044][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:33:13,544][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:33:14,043][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:33:14,541][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:33:15,040][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:33:15,539][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:33:16,039][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:33:16,537][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:33:17,036][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:33:17,534][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:33:18,033][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:33:18,533][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:33:19,032][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:33:19,530][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:33:20,029][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:33:20,528][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:33:21,027][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:33:21,527][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:33:22,027][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:33:22,526][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:33:23,026][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:33:23,524][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:33:24,024][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:33:24,524][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:33:25,026][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:33:25,526][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:33:26,027][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:33:26,527][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:33:27,028][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:33:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:33:28,029][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:33:28,529][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:33:29,028][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:33:29,527][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:33:30,027][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:33:30,526][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:33:31,025][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:33:31,524][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:33:32,023][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:33:32,523][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:33:33,022][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:33:33,523][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:33:34,022][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:33:34,523][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:33:35,023][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:33:35,523][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:33:36,023][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:33:36,523][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:33:37,024][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:33:37,524][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:33:38,025][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:33:38,525][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:33:39,025][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:33:39,525][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:33:40,025][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:33:40,524][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:33:41,023][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:33:41,522][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:33:42,022][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:33:42,521][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:33:43,021][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:33:43,519][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:33:44,017][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 02:33:44,665][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:33:45,377][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:33:45,379][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:33:45,381][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:33:46,063][__main__][INFO] - Iteration 281 took 1m 14s (9.24% Gen, 89.84% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 0m 42s. Estimated total time: 61h 59m 46s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 59s, 500 more iterations: 10h 19m 57s. [2026-03-26 02:33:46,065][__main__][INFO] - Starting iteration 281. [2026-03-26 02:33:46,464][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:33:46,465][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:33:47,239][mllm.models.large_language_model_local][WARNING] - Response Proposed: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:33:49,128][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:33:52,741][__main__][INFO] - Number of regex retries in iteration 281: 2 [2026-03-26 02:33:52,742][__main__][INFO] - agents played in iteration 281 are Alice, Bob [2026-03-26 02:33:53,622][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:33:53,667][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:33:53,668][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:33:53,668][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:33:54,559][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:33:55,016][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:33:55,517][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:33:56,015][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:33:56,514][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:33:57,013][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:33:57,511][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:33:58,009][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:33:58,509][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:33:59,009][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:33:59,509][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:34:00,009][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:34:00,509][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:34:01,009][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:34:01,507][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:34:02,006][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:34:02,506][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:34:03,008][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:34:03,509][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:34:04,010][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:34:04,510][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:34:05,010][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:34:05,509][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:34:06,008][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:34:06,509][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:34:07,008][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:34:07,507][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:34:08,006][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:34:08,505][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:34:09,004][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:34:09,503][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:34:10,002][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:34:10,502][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:34:11,003][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:34:11,502][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:34:12,001][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:34:12,500][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:34:12,999][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:34:13,497][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:34:13,995][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:34:14,494][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:34:14,993][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:34:15,496][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:34:15,997][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:34:16,496][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:34:16,994][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:34:17,493][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:34:17,993][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:34:18,494][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:34:18,993][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:34:19,492][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:34:19,992][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:34:20,490][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:34:20,989][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:34:21,489][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:34:21,988][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:34:22,488][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:34:22,987][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:34:23,486][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:34:23,986][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:34:24,485][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:34:24,984][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:34:25,485][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:34:25,984][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:34:26,483][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:34:26,981][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:34:27,479][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:34:27,978][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:34:28,476][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:34:28,974][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:34:29,472][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:34:29,970][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:34:30,469][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:34:30,968][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:34:31,467][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:34:31,966][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:34:32,465][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:34:32,964][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:34:33,465][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:34:33,964][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:34:34,463][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:34:34,963][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:34:35,463][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:34:35,963][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:34:36,463][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:34:36,964][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:34:37,465][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:34:37,966][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:34:38,465][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:34:38,964][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:34:39,463][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:34:39,961][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:34:40,462][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:34:40,962][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:34:41,461][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:34:41,960][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:34:42,458][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:34:42,957][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:34:43,456][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:34:43,954][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:34:44,453][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:34:44,952][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:34:45,451][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:34:45,951][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:34:46,450][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:34:46,948][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:34:47,447][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:34:47,947][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:34:48,449][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:34:48,948][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:34:49,448][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:34:49,947][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:34:50,446][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:34:50,945][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:34:51,444][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:34:51,944][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:34:52,443][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:34:52,942][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:34:53,441][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:34:53,940][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:34:54,439][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:34:54,937][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:34:55,438][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:34:55,937][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:34:56,436][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:34:56,935][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:34:57,434][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:34:57,933][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:34:58,431][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 02:34:59,072][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:34:59,810][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:34:59,813][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:34:59,814][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:35:00,468][__main__][INFO] - Iteration 282 took 1m 14s (8.48% Gen, 90.63% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 39m 54s. Estimated total time: 61h 40m 13s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 20s, 500 more iterations: 10h 16m 42s. [2026-03-26 02:35:00,470][__main__][INFO] - Starting iteration 282. [2026-03-26 02:35:00,878][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:35:00,879][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:35:07,253][__main__][INFO] - Number of regex retries in iteration 282: 0 [2026-03-26 02:35:07,253][__main__][INFO] - agents played in iteration 282 are Alice, Bob [2026-03-26 02:35:08,127][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:35:08,171][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:35:08,172][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:35:08,173][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:35:08,781][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:35:09,237][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:35:09,743][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:35:10,243][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:35:10,742][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:35:11,242][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:35:11,740][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:35:12,242][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:35:12,742][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:35:13,243][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:35:13,743][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:35:14,242][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:35:14,740][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:35:15,239][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:35:15,741][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:35:16,242][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:35:16,742][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:35:17,243][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:35:17,741][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:35:18,241][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:35:18,739][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:35:19,237][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:35:19,736][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:35:20,234][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:35:20,732][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:35:21,233][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:35:21,732][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:35:22,231][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:35:22,730][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:35:23,230][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:35:23,730][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:35:24,230][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:35:24,729][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:35:25,229][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:35:25,728][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:35:26,227][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:35:26,727][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:35:27,227][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:35:27,727][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:35:28,226][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:35:28,725][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:35:29,223][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:35:29,722][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:35:30,219][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:35:30,717][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:35:31,214][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:35:31,711][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:35:32,210][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:35:32,708][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:35:33,206][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:35:33,705][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:35:34,204][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:35:34,703][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:35:35,204][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:35:35,703][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:35:36,202][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:35:36,701][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:35:37,199][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:35:37,698][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:35:38,197][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:35:38,695][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:35:39,195][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:35:39,694][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:35:40,194][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:35:40,693][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:35:41,192][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:35:41,691][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:35:42,190][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:35:42,690][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:35:43,190][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:35:43,690][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:35:44,189][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:35:44,689][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:35:45,189][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:35:45,692][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:35:46,193][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:35:46,692][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:35:47,192][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:35:47,691][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:35:48,193][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:35:48,693][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:35:49,190][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:35:49,689][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:35:50,189][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:35:50,687][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:35:51,185][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:35:51,683][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:35:52,182][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:35:52,679][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:35:53,177][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:35:53,676][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:35:54,176][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:35:54,676][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:35:55,175][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:35:55,675][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:35:56,175][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:35:56,674][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:35:57,173][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:35:57,671][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:35:58,171][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:35:58,670][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:35:59,170][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:35:59,669][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:36:00,168][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:36:00,667][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:36:01,165][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:36:01,664][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:36:02,163][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:36:02,661][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:36:03,159][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:36:03,657][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:36:04,155][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:36:04,651][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:36:05,150][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:36:05,650][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:36:06,149][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:36:06,649][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:36:07,149][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:36:07,648][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:36:08,147][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:36:08,646][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:36:09,146][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:36:09,646][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:36:10,146][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:36:10,646][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:36:11,146][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:36:11,645][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:36:12,144][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:36:12,643][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 02:36:13,273][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 02:36:14,060][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:36:14,063][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:36:14,064][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:36:14,797][__main__][INFO] - Iteration 283 took 1m 13s (8.62% Gen, 90.38% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 34m 24s. Estimated total time: 61h 35m 56s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 11s, 500 more iterations: 10h 15m 59s. [2026-03-26 02:36:14,799][__main__][INFO] - Starting iteration 283. [2026-03-26 02:36:15,200][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:36:15,200][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:36:21,542][__main__][INFO] - Number of regex retries in iteration 283: 0 [2026-03-26 02:36:21,543][__main__][INFO] - agents played in iteration 283 are Alice, Bob [2026-03-26 02:36:22,415][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:36:22,459][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:36:22,459][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:36:22,460][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:36:23,066][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:36:23,521][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:36:24,025][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:36:24,523][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:36:25,024][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:36:25,523][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:36:26,024][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:36:26,523][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:36:27,022][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:36:27,521][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:36:28,019][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:36:28,517][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:36:29,014][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:36:29,511][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:36:30,010][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:36:30,509][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:36:31,006][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:36:31,507][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:36:32,007][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:36:32,507][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:36:33,007][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:36:33,507][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:36:34,007][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:36:34,507][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:36:35,006][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:36:35,504][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:36:36,002][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:36:36,502][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:36:37,000][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:36:37,499][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:36:37,998][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:36:38,496][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:36:38,994][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:36:39,492][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:36:39,990][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:36:40,489][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:36:40,987][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:36:41,486][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:36:41,985][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:36:42,483][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:36:42,982][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:36:43,482][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:36:43,981][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:36:44,480][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:36:44,978][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:36:45,477][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:36:45,977][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:36:46,476][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:36:46,975][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:36:47,473][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:36:47,972][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:36:48,472][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:36:48,971][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:36:49,469][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:36:49,968][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:36:50,467][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:36:50,965][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:36:51,464][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:36:51,963][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:36:52,462][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:36:52,960][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:36:53,458][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:36:53,958][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:36:54,457][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:36:54,956][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:36:55,455][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:36:55,953][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:36:56,451][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:36:56,949][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:36:57,448][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:36:57,947][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:36:58,446][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:36:58,944][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:36:59,442][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:36:59,940][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:37:00,437][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:37:00,934][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:37:01,432][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:37:01,930][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:37:02,429][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:37:02,927][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:37:03,426][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:37:03,926][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:37:04,425][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:37:04,926][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:37:05,426][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:37:05,926][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:37:06,427][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:37:06,927][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:37:07,425][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:37:07,924][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:37:08,422][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:37:08,920][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:37:09,418][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:37:09,916][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:37:10,414][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:37:10,913][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:37:11,411][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:37:11,910][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:37:12,409][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:37:12,908][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:37:13,407][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:37:13,905][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:37:14,404][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:37:14,903][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:37:15,405][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:37:15,909][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:37:16,409][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:37:16,908][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:37:17,408][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:37:17,908][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:37:18,410][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:37:18,909][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:37:19,407][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:37:19,906][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:37:20,404][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:37:20,904][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:37:21,404][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:37:21,903][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:37:22,400][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:37:22,898][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:37:23,397][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:37:23,896][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:37:24,394][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:37:24,893][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:37:25,392][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:37:25,892][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:37:26,389][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:37:26,888][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 02:37:27,512][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:37:28,256][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:37:28,259][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:37:28,260][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:37:28,909][__main__][INFO] - Iteration 284 took 1m 13s (8.60% Gen, 90.51% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 22m 41s. Estimated total time: 61h 25m 28s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 50s, 500 more iterations: 10h 14m 14s. [2026-03-26 02:37:28,911][__main__][INFO] - Starting iteration 284. [2026-03-26 02:37:29,310][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:37:29,310][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:37:36,316][__main__][INFO] - Number of regex retries in iteration 284: 0 [2026-03-26 02:37:36,317][__main__][INFO] - agents played in iteration 284 are Alice, Bob [2026-03-26 02:37:37,191][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:37:37,235][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:37:37,236][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:37:37,237][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:37:37,844][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:37:38,298][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:37:38,800][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:37:39,298][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:37:39,795][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:37:40,293][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:37:40,790][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:37:41,288][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:37:41,786][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:37:42,284][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:37:42,781][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:37:43,277][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:37:43,774][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:37:44,271][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:37:44,769][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:37:45,266][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:37:45,764][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:37:46,265][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:37:46,764][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:37:47,263][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:37:47,762][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:37:48,261][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:37:48,761][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:37:49,260][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:37:49,758][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:37:50,260][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:37:50,760][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:37:51,260][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:37:51,760][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:37:52,259][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:37:52,760][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:37:53,260][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:37:53,761][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:37:54,260][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:37:54,758][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:37:55,257][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:37:55,756][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:37:56,255][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:37:56,754][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:37:57,253][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:37:57,753][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:37:58,252][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:37:58,750][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:37:59,249][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:37:59,749][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:38:00,249][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:38:00,748][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:38:01,247][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:38:01,747][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:38:02,244][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:38:02,742][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:38:03,240][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:38:03,738][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:38:04,235][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:38:04,732][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:38:05,230][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:38:05,728][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:38:06,227][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:38:06,726][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:38:07,227][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:38:07,726][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:38:08,225][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:38:08,724][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:38:09,223][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:38:09,723][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:38:10,220][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:38:10,718][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:38:11,215][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:38:11,713][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:38:12,211][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:38:12,711][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:38:13,210][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:38:13,709][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:38:14,207][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:38:14,705][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:38:15,205][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:38:15,704][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:38:16,202][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:38:16,700][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:38:17,198][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:38:17,695][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:38:18,195][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:38:18,695][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:38:19,194][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:38:19,692][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:38:20,191][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:38:20,691][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:38:21,190][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:38:21,690][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:38:22,190][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:38:22,690][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:38:23,190][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:38:23,691][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:38:24,192][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:38:24,694][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:38:25,193][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:38:25,694][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:38:26,193][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:38:26,691][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:38:27,191][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:38:27,690][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:38:28,190][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:38:28,689][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:38:29,189][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:38:29,688][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:38:30,186][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:38:30,685][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:38:31,185][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:38:31,685][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:38:32,184][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:38:32,683][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:38:33,183][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:38:33,681][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:38:34,178][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:38:34,676][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:38:35,175][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:38:35,673][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:38:36,172][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:38:36,670][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:38:37,169][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:38:37,666][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:38:38,165][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:38:38,664][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:38:39,164][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:38:39,663][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:38:40,162][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:38:40,661][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:38:41,160][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:38:41,658][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21750 tokens. [2026-03-26 02:38:42,291][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:38:43,038][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:38:43,041][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:38:43,043][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:38:43,693][__main__][INFO] - Iteration 285 took 1m 14s (9.42% Gen, 89.70% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 55h 55m 10s. Estimated total time: 61h 59m 12s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 58s, 500 more iterations: 10h 19m 52s. [2026-03-26 02:38:43,695][__main__][INFO] - Starting iteration 285. [2026-03-26 02:38:44,095][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:38:44,096][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:38:50,676][__main__][INFO] - Number of regex retries in iteration 285: 0 [2026-03-26 02:38:50,677][__main__][INFO] - agents played in iteration 285 are Alice, Bob [2026-03-26 02:38:51,561][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:38:51,605][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:38:51,606][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:38:51,607][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:38:52,206][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:38:52,660][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:38:53,163][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:38:53,662][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:38:54,161][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:38:54,659][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:38:55,157][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:38:55,655][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:38:56,154][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:38:56,652][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:38:57,150][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:38:57,648][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:38:58,146][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:38:58,644][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:38:59,142][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:38:59,640][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:39:00,137][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:39:00,637][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:39:01,136][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:39:01,635][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:39:02,134][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:39:02,634][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:39:03,133][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:39:03,632][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:39:04,131][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:39:04,630][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:39:05,129][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:39:05,628][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:39:06,126][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:39:06,624][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:39:07,123][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:39:07,620][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:39:08,118][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:39:08,616][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:39:09,115][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:39:09,613][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:39:10,113][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:39:10,612][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:39:11,112][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:39:11,611][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:39:12,110][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:39:12,609][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:39:13,108][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:39:13,606][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:39:14,105][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:39:14,604][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:39:15,104][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:39:15,601][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:39:16,099][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:39:16,595][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:39:17,092][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:39:17,589][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:39:18,087][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:39:18,586][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:39:19,083][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:39:19,580][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:39:20,076][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:39:20,576][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:39:21,075][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:39:21,574][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:39:22,073][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:39:22,573][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:39:23,073][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:39:23,573][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:39:24,072][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:39:24,570][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:39:25,069][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:39:25,568][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:39:26,067][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:39:26,567][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:39:27,065][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:39:27,564][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:39:28,063][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:39:28,560][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:39:29,057][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:39:29,555][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:39:30,053][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:39:30,551][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:39:31,049][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:39:31,548][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:39:32,046][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:39:32,545][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:39:33,043][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:39:33,542][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:39:34,041][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:39:34,541][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:39:35,040][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:39:35,540][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:39:36,038][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:39:36,538][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:39:37,036][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:39:37,535][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:39:38,034][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:39:38,534][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:39:39,034][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:39:39,533][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:39:40,032][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:39:40,531][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:39:41,031][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:39:41,530][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:39:42,030][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:39:42,529][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:39:43,029][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:39:43,528][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:39:44,028][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:39:44,528][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:39:45,026][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:39:45,526][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:39:46,025][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:39:46,524][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:39:47,022][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:39:47,520][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:39:48,019][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:39:48,516][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:39:49,013][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:39:49,511][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:39:50,009][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:39:50,506][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:39:51,003][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:39:51,500][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:39:51,996][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:39:52,496][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:39:52,994][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:39:53,493][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:39:53,993][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:39:54,493][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:39:54,993][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:39:55,493][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:39:55,992][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:39:56,616][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 02:39:57,387][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:39:57,390][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:39:57,391][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:39:58,040][__main__][INFO] - Iteration 286 took 1m 13s (8.90% Gen, 90.22% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 32m 1s. Estimated total time: 61h 37m 17s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 14s, 500 more iterations: 10h 16m 12s. [2026-03-26 02:39:58,042][__main__][INFO] - Starting iteration 286. [2026-03-26 02:39:58,441][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:39:58,441][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:40:04,428][__main__][INFO] - Number of regex retries in iteration 286: 0 [2026-03-26 02:40:04,429][__main__][INFO] - agents played in iteration 286 are Alice, Bob [2026-03-26 02:40:05,307][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:40:05,352][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:40:05,352][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:40:05,353][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:40:05,955][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:40:06,409][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:40:06,911][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:40:07,410][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:40:07,909][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:40:08,408][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:40:08,907][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:40:09,405][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:40:09,904][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:40:10,399][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:40:10,893][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:40:11,388][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:40:11,882][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:40:12,377][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:40:12,871][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:40:13,366][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:40:13,860][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:40:14,361][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:40:14,861][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:40:15,362][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:40:15,864][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:40:16,363][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:40:16,862][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:40:17,360][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:40:17,859][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:40:18,360][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:40:18,860][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:40:19,358][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:40:19,857][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:40:20,355][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:40:20,854][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:40:21,353][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:40:21,852][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:40:22,350][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:40:22,847][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:40:23,344][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:40:23,842][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:40:24,338][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:40:24,835][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:40:25,332][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:40:25,830][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:40:26,327][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:40:26,826][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:40:27,860][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:40:28,358][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:40:28,855][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:40:29,353][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:40:29,851][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:40:30,349][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:40:30,848][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:40:31,347][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:40:31,846][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:40:32,346][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:40:32,846][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:40:33,346][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:40:33,846][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:40:34,346][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:40:34,843][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:40:35,340][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:40:35,837][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:40:36,336][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:40:36,833][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:40:37,332][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:40:37,830][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:40:38,329][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:40:38,826][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:40:39,324][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:40:39,821][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:40:40,319][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:40:40,816][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:40:41,313][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:40:41,812][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:40:42,310][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:40:42,805][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:40:43,300][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:40:43,794][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:40:44,288][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:40:44,783][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:40:45,277][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:40:45,771][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:40:46,267][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:40:46,766][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:40:47,267][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:40:47,767][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:40:48,270][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:40:48,769][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:40:49,267][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:40:49,767][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:40:50,267][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:40:50,766][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:40:51,265][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:40:51,765][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:40:52,264][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:40:52,762][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:40:53,260][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:40:53,758][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:40:54,257][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:40:54,754][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:40:55,252][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:40:55,750][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:40:56,248][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:40:56,745][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:40:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:40:57,739][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:40:58,237][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:40:58,734][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:40:59,232][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:40:59,729][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:41:00,227][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:41:00,725][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:41:01,223][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:41:01,719][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:41:02,217][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:41:02,716][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:41:03,214][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:41:03,713][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:41:04,214][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:41:04,714][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:41:05,213][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:41:05,712][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:41:06,212][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:41:06,710][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:41:07,208][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:41:07,705][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:41:08,204][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:41:08,701][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:41:09,198][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:41:09,695][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:41:10,193][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 02:41:10,819][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:41:11,555][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:41:11,559][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:41:11,561][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:41:12,203][__main__][INFO] - Iteration 287 took 1m 13s (8.12% Gen, 91.01% Train). Generation: 5s, Training: 1m 7s. Estimated remaining time: 55h 21m 38s. Estimated total time: 61h 28m 8s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 56s, 500 more iterations: 10h 14m 41s. [2026-03-26 02:41:12,205][__main__][INFO] - Starting iteration 287. [2026-03-26 02:41:12,604][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:41:12,605][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:41:13,176][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 0 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:41:15,424][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:41:19,459][__main__][INFO] - Number of regex retries in iteration 287: 2 [2026-03-26 02:41:19,460][__main__][INFO] - agents played in iteration 287 are Alice, Bob [2026-03-26 02:41:20,334][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:41:20,378][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:41:20,379][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:41:20,380][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:41:20,982][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:41:21,438][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:41:21,944][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:41:22,444][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:41:22,943][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:41:23,441][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:41:23,940][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:41:24,439][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:41:24,938][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:41:25,438][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:41:25,936][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:41:26,434][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:41:26,933][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:41:27,431][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:41:27,932][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:41:28,432][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:41:28,932][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:41:29,433][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:41:29,932][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:41:30,431][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:41:30,930][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:41:31,430][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:41:31,929][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:41:32,428][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:41:32,928][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:41:33,426][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:41:33,925][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:41:34,424][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:41:34,922][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:41:35,420][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:41:35,918][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:41:36,416][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:41:36,915][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:41:37,414][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:41:37,913][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:41:38,413][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:41:38,912][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:41:39,412][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:41:39,912][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:41:40,412][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:41:40,911][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:41:41,409][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:41:41,909][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:41:42,408][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:41:42,906][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:41:43,406][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:41:43,905][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:41:44,403][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:41:44,901][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:41:45,401][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:41:45,901][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:41:46,399][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:41:46,897][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:41:47,395][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:41:47,894][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:41:48,395][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:41:48,893][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:41:49,392][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:41:49,890][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:41:50,389][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:41:50,888][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:41:51,387][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:41:51,887][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:41:52,386][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:41:52,885][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:41:53,383][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:41:53,881][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:41:54,379][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:41:54,878][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:41:55,376][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:41:55,874][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:41:56,373][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:41:56,871][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:41:57,369][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:41:57,867][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:41:58,365][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:41:58,863][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:41:59,360][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:41:59,859][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:42:00,356][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:42:00,854][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:42:01,353][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:42:01,853][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:42:02,352][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:42:02,851][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:42:03,351][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:42:03,851][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:42:04,351][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:42:04,851][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:42:05,350][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:42:05,848][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:42:06,346][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:42:06,844][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:42:07,343][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:42:07,842][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:42:08,339][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:42:08,838][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:42:09,337][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:42:09,835][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:42:10,334][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:42:10,833][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:42:11,333][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:42:11,833][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:42:12,332][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:42:12,831][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:42:13,330][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:42:13,829][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:42:14,328][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:42:14,826][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:42:15,325][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:42:15,825][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:42:16,323][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:42:16,822][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:42:17,319][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:42:17,817][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:42:18,317][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:42:18,814][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:42:19,312][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:42:19,809][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:42:20,307][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:42:20,805][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:42:21,303][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:42:21,802][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:42:22,300][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:42:22,799][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:42:23,296][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:42:23,795][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:42:24,295][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:42:24,794][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 02:42:25,419][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:42:26,705][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:42:26,708][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:42:26,709][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:42:27,710][__main__][INFO] - Iteration 288 took 1m 15s (9.13% Gen, 89.54% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 27m 32s. Estimated total time: 62h 35m 18s. Time estimates for 10 more iterations: 12m 31s, 100 more iterations: 2h 5m 10s, 500 more iterations: 10h 25m 53s. [2026-03-26 02:42:27,712][__main__][INFO] - Starting iteration 288. [2026-03-26 02:42:28,111][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:42:28,112][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:42:29,691][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:42:34,712][__main__][INFO] - Number of regex retries in iteration 288: 1 [2026-03-26 02:42:34,713][__main__][INFO] - agents played in iteration 288 are Alice, Bob [2026-03-26 02:42:35,617][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:42:35,661][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:42:35,662][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:42:35,663][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:42:36,269][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:42:36,723][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:42:37,226][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:42:37,725][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:42:38,223][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:42:38,721][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:42:39,219][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:42:39,716][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:42:40,213][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:42:40,713][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:42:41,212][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:42:41,711][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:42:42,209][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:42:42,707][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:42:43,206][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:42:43,705][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:42:44,203][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:42:44,703][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:42:45,200][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:42:45,697][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:42:46,193][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:42:46,691][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:42:47,188][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:42:47,685][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:42:48,185][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:42:48,703][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:42:49,207][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:42:49,707][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:42:50,206][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:42:50,705][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:42:51,204][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:42:51,703][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:42:52,202][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:42:52,702][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:42:53,201][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:42:53,699][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:42:54,198][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:42:54,696][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:42:55,194][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:42:55,692][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:42:56,190][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:42:56,689][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:42:57,187][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:42:57,686][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:42:58,185][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:42:58,683][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:42:59,180][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:42:59,679][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:43:00,177][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:43:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:43:01,173][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:43:01,671][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:43:02,169][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:43:02,667][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:43:03,165][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:43:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:43:04,159][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:43:04,657][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:43:05,155][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:43:05,653][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:43:06,152][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:43:06,651][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:43:07,149][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:43:07,647][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:43:08,145][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:43:08,642][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:43:09,139][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:43:09,636][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:43:10,134][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:43:10,632][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:43:11,129][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:43:11,627][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:43:12,125][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:43:12,624][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:43:13,123][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:43:13,619][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:43:14,117][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:43:14,616][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:43:15,114][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:43:15,617][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:43:16,117][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:43:16,615][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:43:17,113][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:43:17,610][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:43:18,109][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:43:18,608][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:43:19,106][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:43:19,604][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:43:20,102][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:43:20,601][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:43:21,099][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:43:21,597][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:43:22,095][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:43:22,594][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:43:23,093][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:43:23,591][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:43:24,091][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:43:24,589][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:43:25,087][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:43:25,585][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:43:26,084][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:43:26,583][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:43:27,081][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:43:27,579][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:43:28,078][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:43:28,576][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:43:29,074][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:43:29,574][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:43:30,072][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:43:30,571][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:43:31,070][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:43:31,568][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:43:32,066][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:43:32,565][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:43:33,063][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:43:33,560][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:43:34,058][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:43:34,557][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:43:35,054][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:43:35,553][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:43:36,051][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:43:36,550][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:43:37,048][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:43:37,546][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:43:38,045][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:43:38,543][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:43:39,040][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:43:39,539][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:43:40,037][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 02:43:40,663][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 02:43:41,424][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:43:41,426][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:43:41,427][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:43:42,102][__main__][INFO] - Iteration 289 took 1m 13s (8.92% Gen, 90.16% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 30m 34s. Estimated total time: 61h 39m 34s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 19s, 500 more iterations: 10h 16m 35s. [2026-03-26 02:43:42,104][__main__][INFO] - Starting iteration 289. [2026-03-26 02:43:42,502][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:43:42,503][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:43:43,102][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:43:49,053][__main__][INFO] - Number of regex retries in iteration 289: 1 [2026-03-26 02:43:49,053][__main__][INFO] - agents played in iteration 289 are Alice, Bob [2026-03-26 02:43:49,969][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:43:50,013][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:43:50,014][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:43:50,014][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:43:50,615][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:43:51,069][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:43:51,573][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:43:52,072][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:43:52,570][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:43:53,069][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:43:53,568][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:43:54,067][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:43:54,565][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:43:55,067][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:43:55,567][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:43:56,066][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:43:56,566][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:43:57,065][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:43:57,563][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:43:58,062][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:43:58,560][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:43:59,061][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:43:59,561][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:44:00,061][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:44:00,561][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:44:01,061][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:44:01,560][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:44:02,059][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:44:02,559][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:44:03,059][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:44:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:44:04,057][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:44:04,556][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:44:05,056][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:44:05,555][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:44:06,054][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:44:06,554][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:44:07,053][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:44:07,551][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:44:08,050][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:44:08,548][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:44:09,046][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:44:09,544][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:44:10,042][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:44:10,541][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:44:11,060][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:44:11,564][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:44:12,064][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:44:12,563][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:44:13,064][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:44:13,564][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:44:14,065][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:44:14,565][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:44:15,064][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:44:15,562][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:44:16,060][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:44:16,560][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:44:17,059][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:44:17,558][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:44:18,059][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:44:18,560][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:44:19,058][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:44:19,558][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:44:20,057][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:44:20,555][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:44:21,054][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:44:21,554][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:44:22,053][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:44:22,552][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:44:23,050][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:44:23,548][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:44:24,046][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:44:24,545][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:44:25,043][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:44:25,541][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:44:26,039][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:44:26,537][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:44:27,036][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:44:27,534][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:44:28,032][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:44:28,531][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:44:29,029][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:44:29,528][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:44:30,027][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:44:30,527][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:44:31,027][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:44:31,529][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:44:32,029][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:44:32,530][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:44:33,029][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:44:33,529][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:44:34,028][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:44:34,527][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:44:35,027][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:44:35,528][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:44:36,028][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:44:36,528][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:44:37,028][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:44:37,528][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:44:38,028][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:44:38,528][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:44:39,027][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:44:39,527][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:44:40,028][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:44:40,527][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:44:41,026][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:44:41,524][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:44:42,022][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:44:42,519][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:44:43,018][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:44:43,518][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:44:44,017][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:44:44,518][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:44:45,017][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:44:45,520][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:44:46,021][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:44:46,521][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:44:47,021][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:44:47,521][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:44:48,022][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:44:48,523][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:44:49,022][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:44:49,521][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:44:50,020][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:44:50,519][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:44:51,018][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:44:51,517][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:44:52,015][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:44:52,513][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:44:53,012][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:44:53,511][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:44:54,009][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:44:54,509][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-26 02:44:55,137][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:44:55,903][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:44:55,906][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:44:55,907][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:44:56,648][__main__][INFO] - Iteration 290 took 1m 14s (8.83% Gen, 90.16% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 37m 4s. Estimated total time: 61h 47m 19s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 34s, 500 more iterations: 10h 17m 53s. [2026-03-26 02:44:56,650][__main__][INFO] - Starting iteration 290. [2026-03-26 02:44:57,048][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:44:57,049][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:45:03,898][__main__][INFO] - Number of regex retries in iteration 290: 0 [2026-03-26 02:45:03,899][__main__][INFO] - agents played in iteration 290 are Alice, Bob [2026-03-26 02:45:04,773][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:45:04,817][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:45:04,818][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:45:04,818][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:45:05,422][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:45:05,882][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:45:06,387][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:45:06,886][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:45:07,387][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:45:07,886][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:45:08,385][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:45:08,884][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:45:09,383][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:45:09,884][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:45:10,384][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:45:10,884][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:45:11,386][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:45:11,885][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:45:12,385][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:45:12,884][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:45:13,384][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:45:13,881][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:45:14,380][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:45:14,878][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:45:15,377][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:45:15,875][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:45:16,374][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:45:16,872][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:45:17,370][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:45:17,869][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:45:18,370][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:45:18,869][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:45:19,368][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:45:19,867][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:45:20,365][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:45:20,862][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:45:21,361][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:45:21,858][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:45:22,356][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:45:22,855][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:45:23,353][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:45:23,852][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:45:24,350][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:45:24,849][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:45:25,347][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:45:25,845][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:45:26,344][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:45:26,844][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:45:27,344][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:45:27,844][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:45:28,344][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:45:28,843][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:45:29,342][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:45:29,841][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:45:30,341][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:45:30,840][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:45:31,340][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:45:31,841][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:45:32,341][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:45:32,842][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:45:33,342][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:45:33,844][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:45:34,345][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:45:34,846][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:45:35,346][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:45:35,846][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:45:36,345][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:45:36,845][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:45:37,347][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:45:37,847][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:45:38,347][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:45:38,846][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:45:39,345][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:45:39,843][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:45:40,341][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:45:40,839][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:45:41,338][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:45:41,837][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:45:42,337][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:45:42,837][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:45:43,337][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:45:43,836][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:45:44,336][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:45:44,836][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:45:45,336][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:45:45,835][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:45:46,334][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:45:46,832][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:45:47,332][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:45:47,830][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:45:48,331][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:45:48,831][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:45:49,330][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:45:49,828][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:45:50,326][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:45:50,823][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:45:51,321][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:45:51,818][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:45:52,316][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:45:52,815][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:45:53,313][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:45:53,812][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:45:54,311][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:45:54,809][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:45:55,308][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:45:55,807][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:45:56,305][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:45:56,804][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:45:57,303][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:45:57,801][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:45:58,300][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:45:58,798][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:45:59,296][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:45:59,794][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:46:00,293][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:46:00,791][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:46:01,290][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:46:01,790][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:46:02,289][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:46:02,788][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:46:03,289][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:46:03,788][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:46:04,288][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:46:04,788][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:46:05,289][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:46:05,789][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:46:06,289][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:46:06,789][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:46:07,290][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:46:07,790][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:46:08,290][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:46:08,790][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:46:09,290][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 02:46:09,929][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:46:10,657][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:46:10,660][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:46:10,661][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:46:11,552][__main__][INFO] - Iteration 291 took 1m 14s (9.19% Gen, 89.61% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 53m 41s. Estimated total time: 62h 5m 11s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 51s. [2026-03-26 02:46:11,554][__main__][INFO] - Starting iteration 291. [2026-03-26 02:46:11,955][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:46:11,955][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:46:16,611][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:46:16,765][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:46:17,971][__main__][INFO] - Number of regex retries in iteration 291: 2 [2026-03-26 02:46:17,972][__main__][INFO] - agents played in iteration 291 are Alice, Bob [2026-03-26 02:46:18,869][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:46:18,914][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:46:18,915][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:46:18,915][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:46:19,789][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:46:20,246][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:46:20,751][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:46:21,252][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:46:21,752][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:46:22,253][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:46:22,753][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:46:23,255][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:46:23,755][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:46:24,255][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:46:24,756][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:46:25,257][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:46:25,757][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:46:26,256][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:46:26,756][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:46:27,257][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:46:27,757][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:46:28,256][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:46:28,756][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:46:29,256][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:46:29,755][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:46:30,254][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:46:30,754][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:46:31,254][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:46:31,754][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:46:32,253][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:46:32,752][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:46:33,250][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:46:33,747][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:46:34,245][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:46:34,743][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:46:35,241][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:46:35,739][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:46:36,240][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:46:36,740][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:46:37,239][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:46:37,739][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:46:38,239][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:46:38,739][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:46:39,239][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:46:39,739][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:46:40,238][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:46:40,737][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:46:41,236][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:46:41,735][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:46:42,234][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:46:42,733][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:46:43,234][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:46:43,733][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:46:44,232][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:46:44,731][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:46:45,231][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:46:45,730][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:46:46,230][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:46:46,730][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:46:47,229][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:46:47,728][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:46:48,229][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:46:48,730][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:46:49,228][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:46:49,729][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:46:50,228][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:46:50,727][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:46:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:46:51,724][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:46:52,223][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:46:52,722][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:46:53,221][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:46:53,721][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:46:54,221][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:46:54,720][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:46:55,220][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:46:55,719][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:46:56,218][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:46:56,717][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:46:57,217][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:46:57,716][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:46:58,215][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:46:58,714][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:46:59,213][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:46:59,713][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:47:00,213][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:47:00,713][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:47:01,212][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:47:01,711][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:47:02,210][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:47:02,709][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:47:03,207][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:47:03,707][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:47:04,205][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:47:04,702][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:47:05,199][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:47:05,696][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:47:06,193][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:47:06,691][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:47:07,188][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:47:07,686][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:47:08,185][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:47:08,684][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:47:09,184][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:47:09,685][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:47:10,185][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:47:10,684][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:47:11,183][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:47:11,683][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:47:12,181][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:47:12,679][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:47:13,177][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:47:13,676][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:47:14,176][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:47:14,674][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:47:15,172][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:47:15,670][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:47:16,169][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:47:16,668][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:47:17,167][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:47:17,669][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:47:18,170][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:47:18,669][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:47:19,167][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:47:19,667][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:47:20,164][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:47:20,662][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:47:21,159][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:47:21,658][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:47:22,156][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:47:22,654][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:47:23,154][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:47:23,653][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 02:47:24,277][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 02:47:25,000][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:47:25,002][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:47:25,004][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:47:25,744][__main__][INFO] - Iteration 292 took 1m 13s (8.15% Gen, 90.84% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 16m 45s. Estimated total time: 61h 29m 28s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 58s, 500 more iterations: 10h 14m 54s. [2026-03-26 02:47:25,746][__main__][INFO] - Starting iteration 292. [2026-03-26 02:47:26,145][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:47:26,146][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:47:32,569][__main__][INFO] - Number of regex retries in iteration 292: 0 [2026-03-26 02:47:32,570][__main__][INFO] - agents played in iteration 292 are Alice, Bob [2026-03-26 02:47:33,478][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:47:33,523][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:47:33,524][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:47:33,524][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:47:34,128][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:47:34,583][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:47:35,087][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:47:35,587][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:47:36,086][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:47:36,589][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:47:37,087][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:47:37,585][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:47:38,084][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:47:38,584][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:47:39,083][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:47:39,582][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:47:40,082][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:47:40,580][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:47:41,078][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:47:41,578][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:47:42,080][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:47:42,577][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:47:43,075][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:47:43,574][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:47:44,072][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:47:44,570][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:47:45,067][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:47:45,568][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:47:46,068][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:47:46,569][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:47:47,070][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:47:47,570][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:47:48,069][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:47:48,570][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:47:49,069][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:47:49,568][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:47:50,068][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:47:50,567][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:47:51,067][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:47:51,567][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:47:52,065][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:47:52,564][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:47:53,063][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:47:53,562][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:47:54,063][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:47:54,562][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:47:55,061][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:47:55,561][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:47:56,060][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:47:56,559][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:47:57,059][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:47:57,558][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:47:58,057][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:47:58,558][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:47:59,057][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:47:59,557][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:48:00,056][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:48:00,556][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:48:01,056][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:48:01,556][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:48:02,055][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:48:02,554][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:48:03,054][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:48:03,553][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:48:04,052][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:48:04,551][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:48:05,051][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:48:05,550][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:48:06,050][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:48:06,548][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:48:07,046][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:48:07,544][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:48:08,041][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:48:08,539][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:48:09,036][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:48:09,535][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:48:10,033][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:48:10,531][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:48:11,030][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:48:11,530][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:48:12,030][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:48:12,529][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:48:13,030][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:48:13,530][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:48:14,030][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:48:14,528][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:48:15,025][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:48:15,523][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:48:16,020][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:48:16,517][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:48:17,015][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:48:17,514][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:48:18,012][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:48:18,513][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:48:19,012][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:48:19,511][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:48:20,010][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:48:20,508][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:48:21,009][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:48:21,508][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:48:22,008][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:48:22,508][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:48:23,007][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:48:23,506][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:48:24,005][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:48:24,504][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:48:25,003][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:48:25,502][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:48:26,002][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:48:26,500][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:48:26,999][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:48:27,498][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:48:27,996][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:48:28,499][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:48:28,999][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:48:29,498][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:48:29,998][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:48:30,498][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:48:30,997][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:48:31,497][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:48:31,997][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:48:32,497][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:48:32,998][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:48:33,498][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:48:34,004][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:48:34,505][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:48:35,007][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:48:35,509][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:48:36,011][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:48:36,511][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:48:37,012][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:48:37,513][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:48:38,015][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 02:48:38,700][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 02:48:39,431][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:48:39,433][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:48:39,435][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:48:40,163][__main__][INFO] - Iteration 293 took 1m 14s (8.68% Gen, 90.33% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 26m 58s. Estimated total time: 61h 40m 56s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 21s, 500 more iterations: 10h 16m 49s. [2026-03-26 02:48:40,165][__main__][INFO] - Starting iteration 293. [2026-03-26 02:48:40,566][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:48:40,567][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:48:41,165][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:48:47,115][__main__][INFO] - Number of regex retries in iteration 293: 1 [2026-03-26 02:48:47,115][__main__][INFO] - agents played in iteration 293 are Alice, Bob [2026-03-26 02:48:48,000][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:48:48,045][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:48:48,046][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:48:48,047][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:48:48,650][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:48:49,105][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:48:49,608][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:48:50,107][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:48:50,606][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:48:51,107][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:48:51,606][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:48:52,105][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:48:52,604][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:48:53,103][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:48:53,601][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:48:54,099][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:48:54,597][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:48:55,097][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:48:55,596][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:48:56,095][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:48:56,595][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:48:57,095][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:48:57,594][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:48:58,093][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:48:58,593][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:48:59,092][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:48:59,592][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:49:00,091][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:49:00,591][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:49:01,089][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:49:01,592][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:49:02,091][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:49:02,591][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:49:03,091][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:49:03,589][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:49:04,088][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:49:04,587][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:49:05,085][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:49:05,583][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:49:06,081][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:49:06,580][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:49:07,078][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:49:07,576][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:49:08,074][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:49:08,573][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:49:09,074][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:49:09,574][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:49:10,074][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:49:10,573][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:49:11,072][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:49:11,572][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:49:12,071][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:49:12,571][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:49:13,070][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:49:13,569][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:49:14,068][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:49:14,568][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:49:15,067][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:49:15,570][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:49:16,071][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:49:16,570][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:49:17,068][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:49:17,568][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:49:18,068][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:49:18,569][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:49:19,069][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:49:19,567][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:49:20,065][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:49:20,563][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:49:21,063][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:49:21,562][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:49:22,061][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:49:22,561][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:49:23,059][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:49:23,558][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:49:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:49:24,557][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:49:25,057][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:49:25,556][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:49:26,055][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:49:26,555][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:49:27,054][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:49:27,553][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:49:28,052][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:49:28,551][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:49:29,051][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:49:29,551][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:49:30,050][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:49:30,550][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:49:31,049][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:49:31,548][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:49:32,047][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:49:32,549][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:49:33,048][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:49:33,548][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:49:34,048][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:49:34,548][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:49:35,046][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:49:35,546][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:49:36,046][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:49:36,545][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:49:37,045][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:49:37,543][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:49:38,041][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:49:38,539][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:49:39,038][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:49:39,536][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:49:40,035][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:49:40,535][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:49:41,034][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:49:41,535][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:49:42,035][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:49:42,534][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:49:43,034][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:49:43,534][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:49:44,034][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:49:44,534][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:49:45,032][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:49:45,531][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:49:46,030][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:49:46,530][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:49:47,029][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:49:47,529][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:49:48,030][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:49:48,531][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:49:49,030][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:49:49,528][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:49:50,025][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:49:50,523][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:49:51,019][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:49:51,517][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:49:52,015][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:49:52,512][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 02:49:53,188][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:49:53,934][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:49:53,936][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:49:53,938][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:49:54,968][__main__][INFO] - Iteration 294 took 1m 14s (8.80% Gen, 89.81% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 44m 56s. Estimated total time: 62h 0m 9s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 0s, 500 more iterations: 10h 20m 1s. [2026-03-26 02:49:54,971][__main__][INFO] - Starting iteration 294. [2026-03-26 02:49:55,373][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:49:55,374][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:50:01,167][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:50:01,931][__main__][INFO] - Number of regex retries in iteration 294: 1 [2026-03-26 02:50:01,933][__main__][INFO] - agents played in iteration 294 are Alice, Bob [2026-03-26 02:50:02,804][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:50:03,126][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:50:03,127][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:50:03,127][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:50:03,724][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:50:04,179][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:50:04,686][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:50:05,188][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:50:05,689][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:50:06,190][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:50:06,690][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:50:07,191][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:50:07,691][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:50:08,191][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:50:08,692][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:50:09,192][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:50:09,693][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:50:10,193][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:50:10,693][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:50:11,193][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:50:11,693][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:50:12,194][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:50:12,695][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:50:13,195][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:50:13,696][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:50:14,203][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:50:14,705][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:50:15,206][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:50:15,708][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:50:16,208][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:50:16,709][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:50:17,212][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:50:17,713][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:50:18,215][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:50:18,716][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:50:19,229][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:50:19,731][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:50:20,234][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:50:20,737][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:50:21,239][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:50:21,741][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:50:22,243][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:50:22,746][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:50:23,249][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:50:23,751][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:50:24,256][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:50:24,756][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:50:25,257][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:50:25,758][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:50:26,258][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:50:26,758][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:50:27,258][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:50:27,758][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:50:28,255][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:50:28,752][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:50:29,253][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:50:29,751][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:50:30,249][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:50:30,745][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:50:31,242][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:50:31,739][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:50:32,238][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:50:32,737][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:50:33,235][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:50:33,735][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:50:34,234][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:50:34,733][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:50:35,231][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:50:35,730][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:50:36,230][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:50:36,730][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:50:37,231][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:50:37,731][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:50:38,229][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:50:38,729][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:50:39,228][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:50:39,728][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:50:40,225][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:50:40,723][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:50:41,221][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:50:41,719][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:50:42,217][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:50:42,715][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:50:43,212][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:50:43,709][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:50:44,213][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:50:44,711][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:50:45,211][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:50:45,713][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:50:46,213][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:50:46,712][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:50:47,211][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:50:47,710][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:50:48,210][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:50:48,711][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:50:49,211][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:50:49,709][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:50:50,208][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:50:50,706][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:50:51,204][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:50:51,702][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:50:52,199][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:50:52,699][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:50:53,198][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:50:53,697][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:50:54,196][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:50:54,695][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:50:55,194][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:50:55,694][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:50:56,192][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:50:56,692][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:50:57,191][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:50:57,690][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:50:58,189][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:50:58,687][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:50:59,185][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:50:59,684][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:51:00,179][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:51:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:51:01,172][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:51:01,669][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:51:02,165][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:51:02,661][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:51:03,157][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:51:03,654][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:51:04,154][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:51:04,652][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:51:05,151][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:51:05,651][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:51:06,149][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:51:06,648][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:51:07,146][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:51:07,644][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 02:51:08,323][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:51:09,063][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:51:09,065][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:51:09,067][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:51:09,802][__main__][INFO] - Iteration 295 took 1m 14s (8.81% Gen, 90.20% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 44m 59s. Estimated total time: 62h 1m 27s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 2s, 500 more iterations: 10h 20m 14s. [2026-03-26 02:51:09,804][__main__][INFO] - Starting iteration 295. [2026-03-26 02:51:10,203][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:51:10,204][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:51:17,483][__main__][INFO] - Number of regex retries in iteration 295: 0 [2026-03-26 02:51:17,484][__main__][INFO] - agents played in iteration 295 are Alice, Bob [2026-03-26 02:51:18,358][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:51:18,402][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:51:18,403][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:51:18,404][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:51:19,001][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:51:19,457][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:51:19,961][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:51:20,461][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:51:20,960][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:51:21,460][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:51:21,960][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:51:22,459][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:51:22,959][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:51:23,458][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:51:23,957][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:51:24,456][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:51:24,956][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:51:25,456][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:51:25,955][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:51:26,455][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:51:26,954][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:51:27,455][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:51:27,954][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:51:28,454][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:51:28,953][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:51:29,452][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:51:29,952][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:51:30,454][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:51:30,954][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:51:31,453][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:51:31,952][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:51:32,451][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:51:32,951][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:51:33,450][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:51:33,950][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:51:34,450][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:51:34,948][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:51:35,447][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:51:35,945][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:51:36,444][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:51:36,944][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:51:37,444][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:51:37,944][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:51:38,443][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:51:38,942][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:51:39,440][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:51:39,939][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:51:40,438][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:51:40,937][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:51:41,437][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:51:41,937][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:51:42,435][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:51:42,935][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:51:43,434][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:51:43,933][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:51:44,431][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:51:44,930][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:51:45,429][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:51:45,929][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:51:46,426][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:51:46,924][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:51:47,422][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:51:47,920][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:51:48,419][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:51:48,917][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:51:49,416][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:51:49,914][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:51:50,412][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:51:50,910][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:51:51,411][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:51:51,910][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:51:52,410][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:51:52,910][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:51:53,410][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:51:53,910][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:51:54,411][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:51:54,911][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:51:55,410][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:51:55,912][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:51:56,411][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:51:56,910][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:51:57,410][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:51:57,910][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:51:58,410][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:51:58,909][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:51:59,411][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:51:59,910][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:52:00,409][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:52:00,909][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:52:01,408][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:52:01,907][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:52:02,406][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:52:02,905][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:52:03,404][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:52:03,903][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:52:04,402][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:52:04,901][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:52:05,400][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:52:05,898][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:52:06,396][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:52:06,895][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:52:07,395][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:52:07,895][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:52:08,394][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:52:08,892][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:52:09,392][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:52:09,891][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:52:10,390][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:52:10,891][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:52:11,390][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:52:11,890][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:52:12,390][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:52:12,888][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:52:13,387][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:52:13,886][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:52:14,386][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:52:14,884][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:52:15,385][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:52:15,887][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:52:16,386][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:52:16,885][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:52:17,384][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:52:17,883][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:52:18,383][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:52:18,881][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:52:19,380][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:52:19,878][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:52:20,375][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:52:20,873][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:52:21,371][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:52:21,869][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:52:22,367][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:52:22,865][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 02:52:23,534][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 02:52:24,286][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:52:24,289][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:52:24,290][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:52:25,079][__main__][INFO] - Iteration 296 took 1m 14s (9.72% Gen, 89.22% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 56h 6m 6s. Estimated total time: 62h 23m 49s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 47s, 500 more iterations: 10h 23m 58s. [2026-03-26 02:52:25,082][__main__][INFO] - Starting iteration 296. [2026-03-26 02:52:25,480][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:52:25,481][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:52:31,986][__main__][INFO] - Number of regex retries in iteration 296: 0 [2026-03-26 02:52:31,987][__main__][INFO] - agents played in iteration 296 are Alice, Bob [2026-03-26 02:52:32,907][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:52:32,952][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:52:32,953][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:52:32,953][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:52:33,552][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:52:34,008][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:52:34,511][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:52:35,010][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:52:35,511][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:52:36,010][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:52:36,510][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:52:37,010][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:52:37,509][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:52:38,009][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:52:38,507][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:52:39,008][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:52:39,507][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:52:40,006][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:52:40,506][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:52:41,004][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:52:41,504][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:52:42,002][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:52:42,501][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:52:43,000][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:52:43,498][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:52:43,996][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:52:44,496][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:52:44,996][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:52:45,495][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:52:45,993][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:52:46,492][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:52:46,991][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:52:47,490][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:52:47,990][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:52:48,491][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:52:48,990][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:52:49,488][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:52:49,989][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:52:50,490][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:52:50,990][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:52:51,492][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:52:51,993][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:52:52,493][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:52:52,993][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:52:53,493][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:52:53,991][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:52:54,490][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:52:54,989][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:52:55,488][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:52:55,987][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:52:56,485][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:52:56,983][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:52:57,480][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:52:57,978][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:52:58,477][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:52:58,976][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:52:59,474][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:52:59,972][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:53:00,470][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:53:00,969][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:53:01,468][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:53:01,965][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:53:02,462][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:53:02,960][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:53:03,458][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:53:03,956][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:53:04,453][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:53:04,951][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:53:05,449][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:53:05,947][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:53:06,445][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:53:06,940][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:53:07,439][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:53:07,939][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:53:08,435][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:53:08,935][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:53:09,434][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:53:09,933][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:53:10,432][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:53:10,932][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:53:11,431][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:53:11,931][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:53:12,431][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:53:12,930][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:53:13,430][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:53:13,929][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:53:14,427][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:53:14,925][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:53:15,424][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:53:15,923][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:53:16,421][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:53:16,921][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:53:17,419][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:53:17,918][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:53:18,419][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:53:18,918][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:53:19,417][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:53:19,917][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:53:20,416][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:53:20,916][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:53:21,415][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:53:21,915][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:53:22,415][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:53:22,916][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:53:23,416][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:53:23,916][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:53:24,416][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:53:24,916][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:53:25,416][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:53:25,915][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:53:26,413][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:53:26,912][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:53:27,412][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:53:27,912][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:53:28,411][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:53:28,910][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:53:29,409][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:53:29,908][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:53:30,407][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:53:30,905][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:53:31,403][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:53:31,902][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:53:32,400][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:53:32,898][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:53:33,396][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:53:33,893][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:53:34,392][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:53:34,890][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:53:35,388][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:53:35,886][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:53:36,384][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:53:36,882][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:53:37,379][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21688 tokens. [2026-03-26 02:53:38,000][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:53:38,732][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:53:38,734][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:53:38,736][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:53:39,385][__main__][INFO] - Iteration 297 took 1m 13s (8.80% Gen, 90.32% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 16m 17s. Estimated total time: 61h 35m 15s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 10s, 500 more iterations: 10h 15m 52s. [2026-03-26 02:53:39,387][__main__][INFO] - Starting iteration 297. [2026-03-26 02:53:39,788][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:53:39,789][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:53:46,802][__main__][INFO] - Number of regex retries in iteration 297: 0 [2026-03-26 02:53:46,803][__main__][INFO] - agents played in iteration 297 are Alice, Bob [2026-03-26 02:53:47,680][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:53:47,724][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:53:47,725][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:53:47,726][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:53:48,335][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:53:48,790][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:53:49,293][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:53:49,793][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:53:50,293][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:53:50,792][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:53:51,291][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:53:51,791][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:53:52,291][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:53:52,789][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:53:53,287][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:53:53,786][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:53:54,286][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:53:54,783][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:53:55,283][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:53:55,780][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:53:56,278][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:53:56,776][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:53:57,274][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:53:57,772][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:53:58,269][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:53:58,767][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:53:59,265][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:53:59,762][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:54:00,260][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:54:00,760][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:54:01,259][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:54:01,758][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:54:02,257][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:54:02,757][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:54:03,256][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:54:03,755][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:54:04,255][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:54:04,754][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:54:05,252][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:54:05,750][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:54:06,250][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:54:06,749][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:54:07,248][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:54:07,747][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:54:08,244][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:54:08,744][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:54:09,242][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:54:09,741][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:54:10,240][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:54:10,739][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:54:11,238][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:54:11,737][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:54:12,236][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:54:12,737][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:54:13,236][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:54:13,737][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:54:14,238][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:54:14,738][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:54:15,238][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:54:15,737][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:54:16,237][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:54:16,736][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:54:17,236][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:54:17,735][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:54:18,237][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:54:18,737][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:54:19,236][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:54:19,735][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:54:20,235][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:54:20,735][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:54:21,234][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:54:21,733][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:54:22,233][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:54:22,732][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:54:23,231][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:54:23,731][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:54:24,232][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:54:24,731][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:54:25,230][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:54:25,728][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:54:26,226][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:54:26,723][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:54:27,221][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:54:27,718][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:54:28,216][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:54:28,714][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:54:29,213][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:54:29,711][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:54:30,209][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:54:30,707][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:54:31,205][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:54:31,702][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:54:32,200][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:54:32,698][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:54:33,198][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:54:33,697][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:54:34,197][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:54:34,697][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:54:35,196][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:54:35,695][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:54:36,194][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:54:36,693][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:54:37,192][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:54:37,690][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:54:38,189][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:54:38,687][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:54:39,185][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:54:39,684][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:54:40,182][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:54:40,682][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:54:41,181][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:54:41,679][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:54:42,178][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:54:42,678][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:54:43,177][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:54:43,675][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:54:44,174][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:54:44,674][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:54:45,175][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:54:45,674][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:54:46,174][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:54:46,674][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:54:47,173][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:54:47,673][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:54:48,176][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:54:48,676][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:54:49,175][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:54:49,674][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:54:50,173][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:54:50,671][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:54:51,170][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:54:51,669][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:54:52,170][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 02:54:52,797][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 02:54:53,608][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:54:53,610][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:54:53,612][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:54:54,403][__main__][INFO] - Iteration 298 took 1m 14s (9.40% Gen, 89.54% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 55h 50m 35s. Estimated total time: 62h 10m 47s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 21s, 500 more iterations: 10h 21m 47s. [2026-03-26 02:54:54,406][__main__][INFO] - Starting iteration 298. [2026-03-26 02:54:54,805][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:54:54,806][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:54:55,375][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:55:01,291][__main__][INFO] - Number of regex retries in iteration 298: 1 [2026-03-26 02:55:01,292][__main__][INFO] - agents played in iteration 298 are Alice, Bob [2026-03-26 02:55:02,194][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:55:02,239][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:55:02,240][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:55:02,240][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:55:02,847][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:55:03,303][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:55:03,806][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:55:04,306][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:55:04,804][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:55:05,303][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:55:05,801][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:55:06,300][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:55:06,799][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:55:07,299][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:55:07,799][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:55:08,299][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:55:08,798][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:55:09,297][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:55:09,795][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:55:10,295][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:55:10,794][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:55:11,293][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:55:11,791][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:55:12,290][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:55:12,788][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:55:13,287][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:55:13,785][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:55:14,283][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:55:14,782][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:55:15,281][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:55:15,784][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:55:16,284][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:55:16,784][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:55:17,282][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:55:17,781][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:55:18,282][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:55:18,782][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:55:19,284][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:55:19,784][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:55:20,284][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:55:20,783][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:55:21,282][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:55:21,782][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:55:22,282][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:55:22,782][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:55:23,280][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:55:23,779][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:55:24,277][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:55:24,776][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:55:25,275][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:55:25,773][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:55:26,272][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:55:26,772][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:55:27,270][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:55:27,769][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:55:28,267][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:55:28,765][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:55:29,263][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:55:29,761][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:55:30,259][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:55:30,759][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:55:31,257][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:55:31,755][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:55:32,253][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:55:32,751][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:55:33,248][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:55:33,745][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:55:34,243][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:55:34,741][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:55:35,240][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:55:35,740][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:55:36,239][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:55:36,738][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:55:37,237][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:55:37,737][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:55:38,237][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:55:38,737][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:55:39,236][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:55:39,736][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:55:40,235][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:55:40,736][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:55:41,235][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:55:41,735][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:55:42,234][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:55:42,735][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:55:43,234][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:55:43,733][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:55:44,233][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:55:44,732][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:55:45,233][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:55:45,733][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:55:46,232][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:55:46,731][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:55:47,230][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:55:47,730][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:55:48,233][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:55:48,733][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:55:49,233][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:55:49,732][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:55:50,230][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:55:50,729][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:55:51,228][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:55:51,728][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:55:52,228][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:55:52,728][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:55:53,228][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:55:53,728][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:55:54,228][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:55:54,727][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:55:55,225][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:55:55,722][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:55:56,220][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:55:56,719][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:55:57,217][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:55:57,714][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:55:58,212][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:55:58,710][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:55:59,208][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:55:59,705][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:56:00,202][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:56:00,700][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:56:01,198][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:56:01,696][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:56:02,194][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:56:02,692][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:56:03,189][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:56:03,685][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:56:04,182][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:56:04,680][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:56:05,177][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:56:05,675][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:56:06,172][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:56:06,670][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 02:56:07,301][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 02:56:08,070][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:56:08,073][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:56:08,074][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:56:08,699][__main__][INFO] - Iteration 299 took 1m 13s (8.78% Gen, 90.37% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 13m 18s. Estimated total time: 61h 34m 44s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 9s, 500 more iterations: 10h 15m 47s. [2026-03-26 02:56:08,702][__main__][INFO] - Starting iteration 299. [2026-03-26 02:56:09,111][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:56:09,112][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:56:09,717][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:56:10,785][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:56:10,787][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:56:15,856][__main__][INFO] - Number of regex retries in iteration 299: 3 [2026-03-26 02:56:15,857][__main__][INFO] - agents played in iteration 299 are Alice, Bob [2026-03-26 02:56:16,733][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:56:16,778][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:56:16,778][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:56:16,779][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:56:17,385][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:56:17,842][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:56:18,350][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:56:18,852][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:56:19,353][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:56:19,854][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:56:20,355][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:56:20,856][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:56:21,358][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:56:21,858][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:56:22,357][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:56:22,857][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:56:23,355][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:56:23,854][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:56:24,354][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:56:24,854][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:56:25,355][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:56:25,854][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:56:26,352][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:56:26,850][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:56:27,348][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:56:27,846][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:56:28,344][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:56:28,843][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:56:29,342][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:56:29,841][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:56:30,340][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:56:30,839][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:56:31,338][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:56:31,837][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:56:32,336][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:56:32,836][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:56:33,335][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:56:33,834][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:56:34,329][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:56:34,826][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:56:35,325][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:56:35,823][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:56:36,323][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:56:36,822][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:56:37,321][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:56:37,820][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:56:38,319][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:56:38,819][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:56:39,319][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:56:39,818][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:56:40,318][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:56:40,818][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:56:41,319][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:56:41,817][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:56:42,315][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:56:42,814][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:56:43,313][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:56:43,811][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:56:44,312][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:56:44,811][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:56:45,311][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:56:45,813][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:56:46,313][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:56:46,813][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:56:47,311][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:56:47,812][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:56:48,314][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:56:48,814][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:56:49,312][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:56:49,812][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:56:50,313][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:56:50,813][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:56:51,315][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:56:51,815][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:56:52,316][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:56:52,816][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:56:53,316][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:56:53,816][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:56:54,316][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:56:54,815][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:56:55,314][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:56:55,813][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:56:56,312][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:56:56,810][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:56:57,309][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:56:57,807][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:56:58,306][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:56:58,804][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:56:59,302][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:56:59,800][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:57:00,299][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:57:00,797][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:57:01,295][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:57:01,794][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:57:02,294][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:57:02,794][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:57:03,293][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:57:03,793][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:57:04,292][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:57:04,791][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:57:05,289][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:57:05,787][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:57:06,286][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:57:06,785][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:57:07,284][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:57:07,782][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:57:08,281][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:57:08,780][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:57:09,278][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:57:09,778][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:57:10,278][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:57:10,778][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:57:11,277][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:57:11,777][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:57:12,276][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:57:12,775][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:57:13,275][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:57:13,773][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:57:14,273][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:57:14,772][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:57:15,271][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:57:15,769][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:57:16,267][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:57:16,765][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:57:17,262][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:57:17,761][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:57:18,262][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:57:18,762][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:57:19,261][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:57:19,760][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:57:20,258][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:57:20,757][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:57:21,256][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21689 tokens. [2026-03-26 02:57:21,890][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 02:57:22,706][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:57:22,708][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:57:22,710][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:57:23,550][__main__][INFO] - Iteration 300 took 1m 14s (9.06% Gen, 89.81% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 39m 17s. Estimated total time: 62h 1m 58s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 3s, 500 more iterations: 10h 20m 19s. [2026-03-26 02:57:23,553][__main__][INFO] - Starting iteration 300. [2026-03-26 02:57:23,951][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 5 and human policies 1. [2026-03-26 02:57:23,951][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:57:24,553][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 02:57:30,342][__main__][INFO] - Number of regex retries in iteration 300: 1 [2026-03-26 02:57:30,343][__main__][INFO] - agents played in iteration 300 are Alice, Bob [2026-03-26 02:57:31,217][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:57:31,262][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:57:31,263][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:57:31,263][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:57:31,876][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:57:32,332][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:57:32,837][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:57:33,338][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:57:33,839][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:57:34,339][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:57:34,839][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:57:35,339][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:57:35,840][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:57:36,340][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:57:36,843][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:57:37,343][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:57:37,842][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:57:38,341][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:57:38,841][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:57:39,341][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:57:39,841][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:57:40,339][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:57:40,838][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:57:41,337][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:57:41,836][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:57:42,335][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:57:42,834][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:57:43,333][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:57:43,833][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:57:44,332][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:57:44,830][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:57:45,329][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:57:45,828][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:57:46,326][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:57:46,824][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:57:47,322][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:57:47,821][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:57:48,324][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:57:48,825][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:57:49,324][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:57:49,825][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:57:50,325][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:57:50,825][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:57:51,325][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:57:51,825][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:57:52,323][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:57:52,822][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:57:53,319][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:57:53,819][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:57:54,318][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:57:54,818][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:57:55,316][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:57:55,815][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:57:56,314][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:57:56,812][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:57:57,311][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:57:57,810][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:57:58,310][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:57:58,808][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:57:59,306][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:57:59,804][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:58:00,303][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:58:00,801][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:58:01,299][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:58:01,798][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:58:02,298][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:58:02,797][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:58:03,295][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:58:03,794][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:58:04,294][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:58:04,794][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:58:05,294][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:58:05,794][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:58:06,294][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:58:06,793][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:58:07,293][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:58:07,797][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:58:08,295][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:58:08,794][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:58:09,293][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:58:09,793][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:58:10,292][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:58:10,790][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:58:11,288][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:58:11,787][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:58:12,285][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:58:12,783][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:58:13,282][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:58:13,781][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:58:14,280][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:58:14,778][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:58:15,278][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:58:15,780][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:58:16,280][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:58:16,779][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:58:17,277][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:58:17,776][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:58:18,277][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:58:18,775][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:58:19,274][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:58:19,773][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:58:20,273][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:58:20,775][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:58:21,276][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:58:21,776][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:58:22,277][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:58:22,777][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:58:23,278][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:58:23,779][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:58:24,278][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:58:24,777][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:58:25,276][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:58:25,775][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:58:26,272][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:58:26,771][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:58:27,268][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:58:27,765][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:58:28,263][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:58:28,760][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:58:29,257][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:58:29,755][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:58:30,252][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:58:30,748][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:58:31,245][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:58:31,743][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:58:32,241][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:58:32,740][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:58:33,240][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:58:33,739][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:58:34,239][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:58:34,738][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:58:35,237][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:58:35,736][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 02:58:36,420][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 02:58:37,186][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:58:37,189][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:58:37,190][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:58:39,914][__main__][INFO] - Iteration 301 took 1m 15s (8.41% Gen, 88.00% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 56h 54m 13s. Estimated total time: 63h 18m 11s. Time estimates for 10 more iterations: 12m 39s, 100 more iterations: 2h 6m 36s, 500 more iterations: 10h 33m 1s. [2026-03-26 02:58:39,916][__main__][INFO] - Starting iteration 301. [2026-03-26 02:58:40,316][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 02:58:40,317][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:58:46,724][__main__][INFO] - Number of regex retries in iteration 301: 0 [2026-03-26 02:58:46,725][__main__][INFO] - agents played in iteration 301 are Alice, Bob [2026-03-26 02:58:47,602][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:58:47,646][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 02:58:47,647][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 02:58:47,648][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 02:58:48,253][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 02:58:48,709][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 02:58:49,213][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 02:58:49,712][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 02:58:50,210][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 02:58:50,709][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 02:58:51,207][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 02:58:51,707][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 02:58:52,206][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 02:58:52,706][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 02:58:53,205][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 02:58:53,704][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 02:58:54,203][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 02:58:54,701][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 02:58:55,200][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 02:58:55,699][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 02:58:56,199][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 02:58:56,700][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 02:58:57,200][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 02:58:57,699][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 02:58:58,199][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 02:58:58,698][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 02:58:59,198][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 02:58:59,698][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 02:59:00,198][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 02:59:00,697][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 02:59:01,194][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 02:59:01,693][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 02:59:02,191][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 02:59:02,689][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 02:59:03,187][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 02:59:03,684][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 02:59:04,183][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 02:59:04,681][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 02:59:05,179][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 02:59:05,677][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 02:59:06,176][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 02:59:06,674][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 02:59:07,173][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 02:59:07,672][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 02:59:08,171][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 02:59:08,670][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 02:59:09,169][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 02:59:09,667][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 02:59:10,166][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 02:59:10,663][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 02:59:11,162][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 02:59:11,661][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 02:59:12,160][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 02:59:12,658][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 02:59:13,156][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 02:59:13,653][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 02:59:14,151][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 02:59:14,648][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 02:59:15,146][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 02:59:15,643][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 02:59:16,141][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 02:59:16,641][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 02:59:17,141][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 02:59:17,640][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 02:59:18,141][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 02:59:18,641][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 02:59:19,140][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 02:59:19,639][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 02:59:20,140][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 02:59:20,640][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 02:59:21,138][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 02:59:21,638][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 02:59:22,137][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 02:59:22,637][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 02:59:23,135][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 02:59:23,636][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 02:59:24,137][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 02:59:24,636][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 02:59:25,136][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 02:59:25,636][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 02:59:26,136][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 02:59:26,639][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 02:59:27,139][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 02:59:27,638][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 02:59:28,139][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 02:59:28,638][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 02:59:29,137][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 02:59:29,637][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 02:59:30,137][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 02:59:30,636][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 02:59:31,135][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 02:59:31,634][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 02:59:32,134][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 02:59:32,632][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 02:59:33,130][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 02:59:33,626][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 02:59:34,123][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 02:59:34,620][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 02:59:35,119][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 02:59:35,616][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 02:59:36,113][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 02:59:36,611][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 02:59:37,108][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 02:59:37,606][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 02:59:38,103][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 02:59:38,600][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 02:59:39,098][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 02:59:39,595][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 02:59:40,094][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 02:59:40,593][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 02:59:41,092][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 02:59:41,591][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 02:59:42,090][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 02:59:42,589][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 02:59:43,090][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 02:59:43,590][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 02:59:44,089][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 02:59:44,587][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 02:59:45,086][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 02:59:45,584][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 02:59:46,081][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 02:59:46,579][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 02:59:47,077][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 02:59:47,575][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 02:59:48,073][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 02:59:48,575][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 02:59:49,074][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 02:59:49,574][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 02:59:50,074][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 02:59:50,574][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 02:59:51,074][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 02:59:51,574][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 02:59:52,074][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 02:59:52,756][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 02:59:53,492][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 02:59:53,494][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 02:59:53,496][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 02:59:54,160][__main__][INFO] - Iteration 302 took 1m 13s (8.68% Gen, 90.42% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 7m 4s. Estimated total time: 61h 32m 16s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 22s. [2026-03-26 02:59:54,163][__main__][INFO] - Starting iteration 302. [2026-03-26 02:59:54,562][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 02:59:54,562][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 02:59:57,675][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 11 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:00:00,785][__main__][INFO] - Number of regex retries in iteration 302: 1 [2026-03-26 03:00:00,785][__main__][INFO] - agents played in iteration 302 are Alice, Bob [2026-03-26 03:00:01,658][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:00:01,703][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:00:01,703][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:00:01,704][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:00:02,308][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:00:02,761][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:00:03,264][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:00:03,762][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:00:04,260][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:00:04,758][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:00:05,256][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:00:05,754][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:00:06,251][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:00:06,753][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:00:07,253][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:00:07,752][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:00:08,251][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:00:08,750][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:00:09,249][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:00:09,748][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:00:10,246][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:00:10,751][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:00:11,252][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:00:11,753][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:00:12,253][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:00:12,752][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:00:13,253][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:00:13,754][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:00:14,255][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:00:14,755][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:00:15,254][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:00:15,752][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:00:16,250][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:00:16,749][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:00:17,247][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:00:17,747][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:00:18,249][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:00:18,749][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:00:19,249][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:00:19,749][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:00:20,248][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:00:20,748][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:00:21,249][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:00:21,748][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:00:22,248][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:00:22,747][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:00:23,245][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:00:23,743][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:00:24,241][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:00:24,739][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:00:25,240][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:00:25,739][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:00:26,237][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:00:26,736][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:00:27,235][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:00:27,734][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:00:28,234][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:00:28,733][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:00:29,232][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:00:29,731][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:00:30,230][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:00:30,728][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:00:31,226][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:00:31,724][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:00:32,223][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:00:32,722][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:00:33,221][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:00:33,719][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:00:34,219][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:00:34,717][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:00:35,215][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:00:35,713][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:00:36,211][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:00:36,709][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:00:37,211][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:00:37,709][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:00:38,207][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:00:38,705][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:00:39,203][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:00:39,701][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:00:40,200][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:00:40,699][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:00:41,197][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:00:41,695][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:00:42,194][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:00:42,694][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:00:43,194][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:00:43,693][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:00:44,193][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:00:44,693][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:00:45,193][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:00:45,693][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:00:46,194][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:00:46,694][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:00:47,192][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:00:47,689][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:00:48,188][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:00:48,686][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:00:49,183][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:00:49,682][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:00:50,180][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:00:50,679][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:00:51,178][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:00:51,678][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:00:52,177][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:00:52,677][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:00:53,177][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:00:53,676][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:00:54,176][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:00:54,674][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:00:55,174][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:00:55,673][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:00:56,172][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:00:56,672][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:00:57,171][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:00:57,669][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:00:58,167][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:00:58,665][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:00:59,162][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:00:59,660][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:01:00,158][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:01:00,657][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:01:01,155][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:01:01,653][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:01:02,152][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:01:02,651][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:01:03,149][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:01:03,647][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:01:04,146][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:01:04,645][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:01:05,144][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:01:05,643][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:01:06,141][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 03:01:06,831][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 03:01:07,675][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:01:07,677][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:01:07,679][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:01:08,630][__main__][INFO] - Iteration 303 took 1m 14s (8.40% Gen, 90.31% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 16m 59s. Estimated total time: 61h 43m 25s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 26s, 500 more iterations: 10h 17m 14s. [2026-03-26 03:01:08,632][__main__][INFO] - Starting iteration 303. [2026-03-26 03:01:09,039][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:01:09,039][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:01:10,098][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:01:15,402][__main__][INFO] - Number of regex retries in iteration 303: 1 [2026-03-26 03:01:15,403][__main__][INFO] - agents played in iteration 303 are Alice, Bob [2026-03-26 03:01:16,288][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:01:16,332][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:01:16,333][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:01:16,334][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:01:16,936][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:01:17,391][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:01:17,895][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:01:18,395][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:01:18,894][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:01:19,393][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:01:19,891][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:01:20,391][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:01:20,889][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:01:21,388][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:01:21,886][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:01:22,384][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:01:22,882][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:01:23,380][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:01:23,878][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:01:24,377][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:01:24,875][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:01:25,377][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:01:25,876][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:01:26,376][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:01:26,875][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:01:27,375][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:01:27,874][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:01:28,375][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:01:28,874][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:01:29,373][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:01:29,871][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:01:30,373][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:01:30,874][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:01:31,372][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:01:31,870][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:01:32,368][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:01:32,865][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:01:33,360][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:01:33,854][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:01:34,349][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:01:34,843][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:01:35,337][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:01:35,832][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:01:36,326][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:01:36,820][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:01:37,319][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:01:37,818][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:01:38,317][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:01:38,816][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:01:39,314][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:01:39,813][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:01:40,311][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:01:40,810][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:01:41,308][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:01:41,806][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:01:42,304][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:01:42,801][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:01:43,300][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:01:43,798][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:01:44,296][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:01:44,794][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:01:45,292][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:01:45,790][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:01:46,289][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:01:46,788][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:01:47,286][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:01:47,786][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:01:48,285][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:01:48,786][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:01:49,284][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:01:49,782][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:01:50,280][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:01:50,778][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:01:51,276][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:01:51,776][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:01:52,273][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:01:52,772][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:01:53,270][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:01:53,767][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:01:54,265][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:01:54,763][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:01:55,262][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:01:55,760][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:01:56,258][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:01:56,757][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:01:57,256][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:01:57,755][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:01:58,254][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:01:58,755][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:01:59,254][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:01:59,754][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:02:00,253][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:02:00,753][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:02:01,251][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:02:01,749][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:02:02,248][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:02:02,746][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:02:03,244][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:02:03,742][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:02:04,240][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:02:04,739][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:02:05,235][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:02:05,730][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:02:06,223][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:02:06,718][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:02:07,213][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:02:07,708][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:02:08,202][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:02:08,697][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:02:09,195][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:02:09,695][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:02:10,193][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:02:10,691][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:02:11,190][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:02:11,689][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:02:12,187][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:02:12,685][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:02:13,182][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:02:13,680][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:02:14,178][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:02:14,675][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:02:15,175][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:02:15,673][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:02:16,169][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:02:16,666][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:02:17,164][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:02:17,662][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:02:18,162][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:02:18,662][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:02:19,160][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:02:19,658][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:02:20,158][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:02:20,657][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:02:21,344][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 03:02:22,162][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:02:22,164][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:02:22,166][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:02:22,963][__main__][INFO] - Iteration 304 took 1m 13s (8.61% Gen, 90.31% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 8m 33s. Estimated total time: 61h 36m 14s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 12s, 500 more iterations: 10h 16m 2s. [2026-03-26 03:02:22,967][__main__][INFO] - Starting iteration 304. [2026-03-26 03:02:23,365][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:02:23,366][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:02:23,965][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:02:27,895][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:02:30,060][__main__][INFO] - Number of regex retries in iteration 304: 2 [2026-03-26 03:02:30,061][__main__][INFO] - agents played in iteration 304 are Alice, Bob [2026-03-26 03:02:30,937][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:02:30,981][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:02:30,982][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:02:30,982][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:02:31,588][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:02:32,043][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:02:32,546][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:02:33,046][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:02:33,549][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:02:34,049][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:02:34,549][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:02:35,049][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:02:35,550][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:02:36,047][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:02:36,544][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:02:37,042][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:02:37,542][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:02:38,041][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:02:38,539][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:02:39,038][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:02:39,536][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:02:40,037][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:02:40,538][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:02:41,038][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:02:41,539][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:02:42,039][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:02:42,539][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:02:43,040][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:02:43,547][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:02:44,049][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:02:44,549][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:02:45,050][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:02:45,553][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:02:46,055][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:02:46,555][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:02:47,056][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:02:47,556][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:02:48,060][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:02:48,564][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:02:49,067][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:02:49,572][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:02:50,074][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:02:50,576][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:02:51,077][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:02:51,579][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:02:52,081][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:02:52,582][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:02:53,084][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:02:53,592][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:02:54,094][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:02:54,596][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:02:55,098][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:02:55,600][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:02:56,101][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:02:56,602][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:02:57,103][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:02:57,603][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:02:58,103][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:02:58,609][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:02:59,109][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:02:59,610][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:03:00,109][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:03:00,609][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:03:01,108][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:03:01,608][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:03:02,108][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:03:02,606][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:03:03,106][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:03:03,607][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:03:04,112][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:03:04,612][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:03:05,113][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:03:05,615][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:03:06,115][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:03:06,615][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:03:07,115][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:03:07,616][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:03:08,111][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:03:08,608][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:03:09,104][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:03:09,601][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:03:10,098][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:03:10,595][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:03:11,091][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:03:11,587][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:03:12,086][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:03:12,586][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:03:13,084][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:03:13,583][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:03:14,082][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:03:14,582][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:03:15,080][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:03:15,579][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:03:16,076][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:03:16,574][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:03:17,073][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:03:17,572][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:03:18,076][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:03:18,575][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:03:19,075][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:03:19,574][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:03:20,073][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:03:20,571][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:03:21,069][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:03:21,568][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:03:22,065][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:03:22,563][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:03:23,062][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:03:23,562][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:03:24,062][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:03:24,562][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:03:25,063][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:03:25,563][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:03:26,063][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:03:26,562][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:03:27,061][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:03:27,560][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:03:28,060][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:03:28,560][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:03:29,059][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:03:29,558][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:03:30,057][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:03:30,557][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:03:31,056][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:03:31,555][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:03:32,054][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:03:32,554][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:03:33,055][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:03:33,554][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:03:34,054][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:03:34,553][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:03:35,051][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:03:35,550][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:03:36,229][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 03:03:37,006][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:03:37,009][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:03:37,010][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:03:37,802][__main__][INFO] - Iteration 305 took 1m 14s (8.99% Gen, 89.94% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 32m 56s. Estimated total time: 62h 1m 51s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 3s, 500 more iterations: 10h 20m 18s. [2026-03-26 03:03:37,804][__main__][INFO] - Starting iteration 305. [2026-03-26 03:03:38,213][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:03:38,213][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:03:44,186][__main__][INFO] - Number of regex retries in iteration 305: 0 [2026-03-26 03:03:44,468][__main__][INFO] - agents played in iteration 305 are Alice, Bob [2026-03-26 03:03:45,339][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:03:45,383][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:03:45,384][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:03:45,385][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:03:46,001][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:03:46,455][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:03:46,959][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:03:47,458][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:03:47,958][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:03:48,459][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:03:48,958][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:03:49,459][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:03:49,958][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:03:50,459][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:03:50,962][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:03:51,462][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:03:51,962][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:03:52,462][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:03:52,962][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:03:53,462][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:03:53,963][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:03:54,462][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:03:54,960][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:03:55,459][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:03:55,957][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:03:56,455][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:03:56,954][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:03:57,452][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:03:57,949][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:03:58,448][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:03:58,948][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:03:59,448][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:03:59,948][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:04:00,448][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:04:00,948][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:04:01,447][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:04:01,946][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:04:02,447][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:04:02,949][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:04:03,452][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:04:03,955][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:04:04,457][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:04:04,958][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:04:05,460][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:04:05,962][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:04:06,464][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:04:06,965][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:04:07,465][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:04:07,966][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:04:08,474][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:04:08,976][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:04:09,478][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:04:09,980][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:04:10,481][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:04:10,982][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:04:11,482][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:04:11,983][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:04:12,483][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:04:12,984][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:04:13,489][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:04:13,991][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:04:14,494][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:04:14,996][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:04:15,502][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:04:16,006][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:04:16,508][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:04:17,010][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:04:17,511][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:04:18,015][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:04:18,527][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:04:19,029][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:04:19,530][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:04:20,032][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:04:20,533][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:04:21,033][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:04:21,534][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:04:22,036][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:04:22,538][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:04:23,039][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:04:23,550][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:04:24,051][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:04:24,552][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:04:25,056][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:04:25,558][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:04:26,059][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:04:26,559][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:04:27,058][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:04:27,558][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:04:28,058][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:04:28,559][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:04:29,058][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:04:29,558][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:04:30,057][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:04:30,555][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:04:31,053][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:04:31,552][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:04:32,051][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:04:32,550][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:04:33,049][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:04:33,548][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:04:34,047][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:04:34,547][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:04:35,046][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:04:35,546][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:04:36,048][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:04:36,548][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:04:37,048][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:04:37,549][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:04:38,049][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:04:38,548][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:04:39,048][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:04:39,556][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:04:40,057][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:04:40,557][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:04:41,056][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:04:41,556][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:04:42,056][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:04:42,556][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:04:43,056][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:04:43,556][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:04:44,058][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:04:44,557][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:04:45,056][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:04:45,555][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:04:46,055][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:04:46,556][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:04:47,056][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:04:47,556][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:04:48,057][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:04:48,560][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:04:49,061][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:04:49,561][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:04:50,061][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 03:04:50,755][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:04:51,541][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:04:51,543][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:04:51,545][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:04:52,364][__main__][INFO] - Iteration 306 took 1m 14s (8.43% Gen, 90.46% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 17m 26s. Estimated total time: 61h 47m 36s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 56s. [2026-03-26 03:04:52,367][__main__][INFO] - Starting iteration 306. [2026-03-26 03:04:52,766][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:04:52,767][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:04:59,258][__main__][INFO] - Number of regex retries in iteration 306: 0 [2026-03-26 03:04:59,259][__main__][INFO] - agents played in iteration 306 are Alice, Bob [2026-03-26 03:05:00,140][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:05:00,185][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:05:00,185][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:05:00,186][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:05:00,786][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:05:01,241][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:05:01,744][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:05:02,244][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:05:02,744][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:05:03,243][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:05:03,741][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:05:04,240][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:05:04,738][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:05:05,238][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:05:05,736][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:05:06,235][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:05:06,734][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:05:07,231][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:05:07,729][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:05:08,227][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:05:08,725][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:05:09,225][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:05:09,723][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:05:10,224][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:05:10,723][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:05:11,220][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:05:11,718][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:05:12,218][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:05:12,716][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:05:13,218][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:05:13,716][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:05:14,215][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:05:14,713][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:05:15,211][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:05:15,709][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:05:16,207][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:05:16,706][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:05:17,204][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:05:17,703][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:05:18,204][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:05:18,703][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:05:19,201][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:05:19,699][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:05:20,196][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:05:20,694][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:05:21,197][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:05:21,697][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:05:22,196][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:05:22,695][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:05:23,195][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:05:23,694][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:05:24,193][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:05:24,693][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:05:25,194][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:05:25,694][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:05:26,194][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:05:26,693][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:05:27,192][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:05:27,691][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:05:28,190][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:05:28,689][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:05:29,186][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:05:29,683][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:05:30,181][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:05:30,678][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:05:31,175][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:05:31,672][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:05:32,168][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:05:32,665][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:05:33,163][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:05:33,661][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:05:34,159][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:05:34,657][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:05:35,154][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:05:35,652][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:05:36,149][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:05:36,646][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:05:37,144][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:05:37,643][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:05:38,142][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:05:38,646][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:05:39,145][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:05:39,644][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:05:40,143][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:05:40,642][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:05:41,143][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:05:41,644][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:05:42,145][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:05:42,644][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:05:43,144][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:05:43,649][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:05:44,150][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:05:44,650][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:05:45,150][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:05:45,655][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:05:46,156][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:05:46,657][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:05:47,157][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:05:47,658][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:05:48,160][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:05:48,667][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:05:49,168][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:05:49,671][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:05:50,172][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:05:50,674][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:05:51,175][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:05:51,676][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:05:52,177][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:05:52,679][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:05:53,180][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:05:53,686][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:05:54,190][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:05:54,692][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:05:55,194][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:05:55,696][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:05:56,198][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:05:56,700][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:05:57,201][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:05:57,701][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:05:58,202][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:05:58,711][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:05:59,213][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:05:59,715][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:06:00,217][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:06:00,718][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:06:01,219][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:06:01,718][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:06:02,220][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:06:02,720][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:06:03,219][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:06:03,726][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:06:04,225][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:06:04,724][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:06:05,362][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.16%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:06:06,138][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:06:06,140][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:06:06,142][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:06:06,802][__main__][INFO] - Iteration 307 took 1m 14s (8.77% Gen, 90.34% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 10m 22s. Estimated total time: 61h 41m 47s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 23s, 500 more iterations: 10h 16m 57s. [2026-03-26 03:06:06,804][__main__][INFO] - Starting iteration 307. [2026-03-26 03:06:07,204][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:06:07,205][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:06:07,797][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:06:13,741][__main__][INFO] - Number of regex retries in iteration 307: 1 [2026-03-26 03:06:14,021][__main__][INFO] - agents played in iteration 307 are Alice, Bob [2026-03-26 03:06:14,915][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:06:14,960][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:06:14,961][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:06:14,961][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:06:15,565][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:06:16,020][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:06:16,523][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:06:17,022][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:06:17,521][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:06:18,021][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:06:18,522][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:06:19,022][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:06:19,521][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:06:20,019][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:06:20,516][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:06:21,014][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:06:21,512][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:06:22,009][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:06:22,505][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:06:23,003][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:06:23,501][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:06:24,000][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:06:24,498][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:06:24,996][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:06:25,497][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:06:25,996][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:06:26,493][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:06:26,992][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:06:27,490][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:06:27,988][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:06:28,486][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:06:28,985][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:06:29,483][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:06:29,981][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:06:30,480][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:06:30,979][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:06:31,478][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:06:31,978][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:06:32,477][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:06:32,976][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:06:33,475][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:06:33,974][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:06:34,472][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:06:34,970][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:06:35,468][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:06:35,969][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:06:36,469][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:06:36,969][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:06:37,467][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:06:37,967][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:06:38,467][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:06:38,967][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:06:39,466][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:06:39,965][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:06:40,463][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:06:40,962][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:06:41,459][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:06:41,957][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:06:42,456][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:06:42,955][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:06:43,453][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:06:43,952][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:06:44,451][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:06:44,950][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:06:45,449][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:06:45,947][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:06:46,446][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:06:46,945][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:06:47,445][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:06:47,945][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:06:48,445][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:06:48,944][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:06:49,443][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:06:49,941][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:06:50,440][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:06:50,938][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:06:51,437][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:06:51,935][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:06:52,434][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:06:52,931][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:06:53,429][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:06:53,926][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:06:54,423][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:06:54,921][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:06:55,419][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:06:55,918][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:06:56,416][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:06:56,914][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:06:57,412][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:06:57,910][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:06:58,408][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:06:58,905][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:06:59,404][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:06:59,902][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:07:00,401][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:07:00,900][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:07:01,398][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:07:01,897][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:07:02,395][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:07:02,894][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:07:03,394][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:07:03,893][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:07:04,397][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:07:04,897][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:07:05,396][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:07:05,895][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:07:06,394][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:07:06,895][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:07:07,397][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:07:07,897][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:07:08,397][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:07:08,897][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:07:09,403][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:07:09,904][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:07:10,406][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:07:10,908][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:07:11,409][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:07:11,908][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:07:12,407][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:07:12,906][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:07:13,405][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:07:13,905][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:07:14,412][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:07:14,912][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:07:15,415][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:07:15,919][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:07:16,421][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:07:16,922][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:07:17,423][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:07:17,925][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:07:18,427][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:07:18,928][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:07:19,435][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:07:20,110][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 03:07:21,122][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:07:21,124][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:07:21,126][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:07:21,881][__main__][INFO] - Iteration 308 took 1m 14s (9.13% Gen, 89.86% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 41m 14s. Estimated total time: 62h 13m 54s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 27s, 500 more iterations: 10h 22m 19s. [2026-03-26 03:07:21,885][__main__][INFO] - Starting iteration 308. [2026-03-26 03:07:22,288][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:07:22,289][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:07:24,063][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:07:29,091][__main__][INFO] - Number of regex retries in iteration 308: 1 [2026-03-26 03:07:29,092][__main__][INFO] - agents played in iteration 308 are Alice, Bob [2026-03-26 03:07:29,970][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:07:30,015][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:07:30,016][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:07:30,016][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:07:30,621][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:07:31,075][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:07:31,583][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:07:32,082][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:07:32,581][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:07:33,080][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:07:33,579][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:07:34,078][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:07:34,577][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:07:35,075][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:07:35,573][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:07:36,070][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:07:36,570][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:07:37,068][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:07:37,565][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:07:38,065][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:07:38,564][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:07:39,064][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:07:39,564][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:07:40,064][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:07:40,564][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:07:41,063][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:07:41,563][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:07:42,062][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:07:42,562][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:07:43,062][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:07:43,562][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:07:44,062][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:07:44,562][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:07:45,061][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:07:45,560][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:07:46,059][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:07:46,559][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:07:47,059][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:07:47,559][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:07:48,060][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:07:48,563][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:07:49,063][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:07:49,563][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:07:50,064][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:07:50,565][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:07:51,067][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:07:51,567][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:07:52,068][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:07:52,567][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:07:53,065][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:07:53,565][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:07:54,065][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:07:54,565][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:07:55,067][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:07:55,568][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:07:56,068][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:07:56,567][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:07:57,067][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:07:57,567][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:07:58,067][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:07:58,566][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:07:59,064][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:07:59,562][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:08:00,060][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:08:00,558][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:08:01,056][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:08:01,553][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:08:02,051][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:08:02,550][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:08:03,048][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:08:03,547][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:08:04,045][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:08:04,544][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:08:05,042][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:08:05,541][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:08:06,040][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:08:06,538][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:08:07,037][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:08:07,536][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:08:08,035][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:08:08,532][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:08:09,030][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:08:09,527][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:08:10,025][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:08:10,523][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:08:11,023][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:08:11,523][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:08:12,022][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:08:12,521][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:08:13,020][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:08:13,520][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:08:14,019][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:08:14,518][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:08:15,017][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:08:15,516][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:08:16,015][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:08:16,513][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:08:17,011][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:08:17,509][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:08:18,008][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:08:18,508][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:08:19,008][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:08:19,508][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:08:20,008][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:08:20,508][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:08:21,007][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:08:21,505][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:08:22,005][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:08:22,504][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:08:23,003][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:08:23,501][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:08:24,000][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:08:24,499][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:08:24,998][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:08:25,498][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:08:25,998][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:08:26,497][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:08:26,997][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:08:27,496][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:08:27,994][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:08:28,493][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:08:28,992][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:08:29,490][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:08:29,990][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:08:30,488][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:08:30,986][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:08:31,484][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:08:31,982][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:08:32,480][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:08:32,978][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:08:33,478][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:08:33,977][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:08:34,477][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:08:35,145][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 03:08:35,883][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:08:35,886][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:08:35,888][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:08:36,607][__main__][INFO] - Iteration 309 took 1m 14s (9.15% Gen, 89.88% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 22m 6s. Estimated total time: 61h 56m 0s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 52s, 500 more iterations: 10h 19m 20s. [2026-03-26 03:08:36,609][__main__][INFO] - Starting iteration 309. [2026-03-26 03:08:37,015][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:08:37,016][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:08:43,441][__main__][INFO] - Number of regex retries in iteration 309: 0 [2026-03-26 03:08:43,442][__main__][INFO] - agents played in iteration 309 are Alice, Bob [2026-03-26 03:08:44,330][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:08:44,653][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:08:44,654][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:08:44,654][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:08:45,257][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:08:45,713][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:08:46,219][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:08:46,719][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:08:47,219][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:08:47,719][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:08:48,221][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:08:48,722][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:08:49,221][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:08:49,721][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:08:50,220][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:08:50,720][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:08:51,220][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:08:51,723][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:08:52,223][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:08:52,724][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:08:53,223][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:08:53,728][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:08:54,230][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:08:54,731][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:08:55,235][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:08:55,737][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:08:56,239][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:08:56,746][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:08:57,249][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:08:57,751][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:08:58,253][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:08:58,756][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:08:59,258][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:08:59,760][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:09:00,262][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:09:00,764][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:09:01,266][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:09:01,771][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:09:02,273][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:09:02,776][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:09:03,278][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:09:03,780][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:09:04,282][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:09:04,784][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:09:05,286][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:09:05,787][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:09:06,287][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:09:06,794][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:09:07,295][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:09:07,797][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:09:08,557][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:09:09,108][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:09:09,608][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:09:10,108][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:09:10,607][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:09:11,107][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:09:11,606][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:09:12,113][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:09:12,612][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:09:13,112][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:09:13,612][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:09:14,112][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:09:14,611][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:09:15,110][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:09:15,609][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:09:16,107][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:09:16,680][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:09:17,184][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:09:17,683][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:09:18,185][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:09:18,684][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:09:19,184][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:09:19,683][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:09:20,181][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:09:20,680][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:09:21,178][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:09:21,678][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:09:22,178][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:09:22,677][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:09:23,176][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:09:23,677][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:09:24,177][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:09:24,677][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:09:25,177][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:09:25,677][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:09:26,177][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:09:26,678][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:09:27,179][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:09:27,680][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:09:28,180][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:09:28,680][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:09:29,180][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:09:29,681][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:09:30,181][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:09:30,681][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:09:31,180][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:09:31,685][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:09:32,184][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:09:32,684][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:09:33,183][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:09:33,684][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:09:34,184][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:09:34,684][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:09:35,184][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:09:35,685][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:09:36,185][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:09:36,685][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:09:37,186][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:09:37,687][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:09:38,186][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:09:38,684][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:09:39,182][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:09:39,680][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:09:40,179][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:09:40,678][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:09:41,176][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:09:41,677][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:09:42,175][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:09:42,674][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:09:43,171][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:09:43,669][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:09:44,168][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:09:44,666][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:09:45,164][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:09:45,664][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:09:46,162][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:09:46,660][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:09:47,159][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:09:47,657][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:09:48,156][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:09:48,655][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:09:49,155][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:09:49,655][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 03:09:50,342][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:05 [2026-03-26 03:09:50,993][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:09:50,995][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:09:50,997][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:09:51,749][__main__][INFO] - Iteration 310 took 1m 14s (8.60% Gen, 90.39% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 41m 35s. Estimated total time: 62h 16m 44s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 33s, 500 more iterations: 10h 22m 47s. [2026-03-26 03:09:51,753][__main__][INFO] - Starting iteration 310. [2026-03-26 03:09:52,152][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:09:52,153][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:09:58,887][__main__][INFO] - Number of regex retries in iteration 310: 0 [2026-03-26 03:09:58,888][__main__][INFO] - agents played in iteration 310 are Alice, Bob [2026-03-26 03:09:59,790][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:09:59,834][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:09:59,835][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:09:59,836][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:10:00,441][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:10:00,896][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:10:01,399][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:10:01,899][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:10:02,398][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:10:02,898][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:10:03,398][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:10:03,897][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:10:04,396][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:10:04,894][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:10:05,394][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:10:05,895][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:10:06,395][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:10:06,895][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:10:07,393][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:10:07,892][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:10:08,390][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:10:08,889][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:10:09,387][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:10:09,887][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:10:10,386][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:10:10,884][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:10:11,383][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:10:11,881][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:10:12,380][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:10:12,880][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:10:13,379][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:10:13,878][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:10:14,378][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:10:14,877][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:10:15,378][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:10:15,880][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:10:16,380][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:10:16,881][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:10:17,381][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:10:17,882][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:10:18,384][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:10:18,883][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:10:19,383][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:10:19,883][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:10:20,383][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:10:20,881][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:10:21,380][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:10:21,878][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:10:22,376][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:10:22,874][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:10:23,372][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:10:23,870][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:10:24,368][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:10:24,867][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:10:25,368][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:10:25,866][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:10:26,366][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:10:26,865][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:10:27,365][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:10:27,864][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:10:28,364][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:10:28,863][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:10:29,362][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:10:29,862][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:10:30,361][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:10:30,860][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:10:31,359][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:10:31,859][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:10:32,358][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:10:32,858][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:10:33,357][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:10:33,856][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:10:34,355][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:10:34,854][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:10:35,353][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:10:35,852][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:10:36,352][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:10:36,851][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:10:37,350][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:10:37,849][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:10:38,349][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:10:38,847][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:10:39,346][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:10:39,846][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:10:40,346][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:10:40,845][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:10:41,344][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:10:41,843][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:10:42,342][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:10:42,841][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:10:43,339][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:10:43,838][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:10:44,337][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:10:44,837][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:10:45,336][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:10:45,834][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:10:46,333][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:10:46,837][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:10:47,339][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:10:47,839][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:10:48,341][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:10:48,843][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:10:49,345][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:10:49,847][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:10:50,348][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:10:50,849][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:10:51,350][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:10:51,855][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:10:52,358][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:10:52,860][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:10:53,360][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:10:53,861][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:10:54,362][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:10:54,861][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:10:55,361][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:10:55,861][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:10:56,360][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:10:56,866][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:10:57,369][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:10:57,871][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:10:58,372][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:10:58,873][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:10:59,374][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:10:59,875][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:11:00,377][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:11:00,878][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:11:01,380][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:11:01,883][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:11:02,384][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:11:02,886][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:11:03,387][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:11:03,888][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:11:04,390][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 03:11:05,092][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:11:05,898][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:11:05,901][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:11:05,902][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:11:06,520][__main__][INFO] - Iteration 311 took 1m 14s (9.06% Gen, 90.11% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 21m 59s. Estimated total time: 61h 58m 23s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 56s, 500 more iterations: 10h 19m 43s. [2026-03-26 03:11:06,522][__main__][INFO] - Starting iteration 311. [2026-03-26 03:11:06,921][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:11:06,922][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:11:13,481][__main__][INFO] - Number of regex retries in iteration 311: 0 [2026-03-26 03:11:13,482][__main__][INFO] - agents played in iteration 311 are Alice, Bob [2026-03-26 03:11:14,376][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:11:14,421][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:11:14,422][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:11:14,422][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:11:15,027][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:11:15,482][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:11:15,985][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:11:16,484][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:11:16,982][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:11:17,481][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:11:17,980][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:11:18,482][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:11:18,981][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:11:19,482][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:11:19,982][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:11:20,481][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:11:20,981][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:11:21,481][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:11:21,981][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:11:22,481][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:11:22,981][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:11:23,480][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:11:23,979][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:11:24,479][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:11:24,979][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:11:25,479][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:11:25,978][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:11:26,477][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:11:26,977][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:11:27,475][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:11:27,973][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:11:28,472][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:11:28,970][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:11:29,469][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:11:29,967][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:11:30,467][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:11:30,966][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:11:31,464][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:11:31,962][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:11:32,461][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:11:32,963][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:11:33,462][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:11:33,961][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:11:34,460][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:11:34,959][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:11:35,457][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:11:35,955][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:11:36,454][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:11:36,952][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:11:37,452][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:11:37,951][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:11:38,450][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:11:38,948][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:11:39,447][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:11:39,945][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:11:40,444][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:11:40,943][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:11:41,443][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:11:41,941][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:11:42,440][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:11:42,939][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:11:43,437][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:11:43,937][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:11:44,437][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:11:44,936][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:11:45,438][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:11:45,940][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:11:46,440][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:11:46,939][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:11:47,438][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:11:47,939][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:11:48,440][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:11:48,939][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:11:49,437][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:11:49,936][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:11:50,434][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:11:50,932][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:11:51,431][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:11:51,931][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:11:52,430][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:11:52,929][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:11:53,429][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:11:53,928][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:11:54,428][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:11:54,928][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:11:55,427][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:11:55,926][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:11:56,426][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:11:56,925][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:11:57,425][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:11:57,924][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:11:58,423][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:11:58,923][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:11:59,422][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:11:59,921][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:12:00,420][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:12:00,919][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:12:01,417][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:12:01,915][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:12:02,414][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:12:02,912][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:12:03,410][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:12:03,909][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:12:04,408][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:12:04,907][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:12:05,406][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:12:05,905][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:12:06,404][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:12:06,903][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:12:07,402][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:12:07,901][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:12:08,400][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:12:08,899][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:12:09,398][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:12:09,897][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:12:10,395][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:12:10,894][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:12:11,392][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:12:11,890][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:12:12,390][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:12:12,890][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:12:13,389][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:12:13,888][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:12:14,387][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:12:14,886][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:12:15,385][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:12:15,885][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:12:16,386][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:12:16,887][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:12:17,387][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:12:17,888][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:12:18,389][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:12:18,890][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:12:19,550][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 03:12:20,302][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:12:20,304][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:12:20,306][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:12:21,032][__main__][INFO] - Iteration 312 took 1m 14s (8.85% Gen, 90.17% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 7m 55s. Estimated total time: 61h 45m 34s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 31s, 500 more iterations: 10h 17m 35s. [2026-03-26 03:12:21,034][__main__][INFO] - Starting iteration 312. [2026-03-26 03:12:21,444][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:12:21,444][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:12:27,738][__main__][INFO] - Number of regex retries in iteration 312: 0 [2026-03-26 03:12:27,739][__main__][INFO] - agents played in iteration 312 are Alice, Bob [2026-03-26 03:12:28,621][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:12:28,666][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:12:28,666][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:12:28,667][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:12:29,273][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:12:29,725][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:12:30,227][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:12:30,724][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:12:31,221][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:12:31,719][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:12:32,217][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:12:32,714][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:12:33,211][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:12:33,711][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:12:34,209][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:12:34,709][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:12:35,207][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:12:35,706][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:12:36,205][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:12:36,705][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:12:37,204][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:12:37,702][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:12:38,202][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:12:38,701][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:12:39,200][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:12:39,699][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:12:40,197][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:12:40,696][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:12:41,195][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:12:41,692][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:12:42,189][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:12:42,687][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:12:43,185][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:12:43,684][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:12:44,181][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:12:44,678][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:12:45,175][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:12:45,674][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:12:46,172][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:12:46,672][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:12:47,171][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:12:47,669][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:12:48,172][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:12:48,672][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:12:49,171][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:12:49,671][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:12:50,171][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:12:50,669][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:12:51,168][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:12:51,668][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:12:52,166][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:12:52,666][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:12:53,164][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:12:53,663][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:12:54,162][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:12:54,661][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:12:55,161][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:12:55,661][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:12:56,159][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:12:56,657][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:12:57,156][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:12:57,654][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:12:58,154][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:12:58,655][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:12:59,156][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:12:59,655][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:13:00,155][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:13:00,653][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:13:01,152][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:13:01,650][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:13:02,148][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:13:02,646][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:13:03,144][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:13:03,642][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:13:04,140][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:13:04,637][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:13:05,135][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:13:05,633][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:13:06,133][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:13:06,633][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:13:07,133][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:13:07,632][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:13:08,131][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:13:08,630][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:13:09,129][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:13:09,628][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:13:10,127][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:13:10,627][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:13:11,127][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:13:11,626][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:13:12,125][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:13:12,624][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:13:13,123][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:13:13,620][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:13:14,119][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:13:14,618][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:13:15,115][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:13:15,616][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:13:16,114][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:13:16,612][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:13:17,108][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:13:17,607][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:13:18,107][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:13:18,609][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:13:19,109][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:13:19,607][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:13:20,106][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:13:20,605][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:13:21,103][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:13:21,603][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:13:22,104][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:13:22,604][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:13:23,104][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:13:23,603][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:13:24,103][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:13:24,602][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:13:25,103][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:13:25,603][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:13:26,101][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:13:26,599][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:13:27,097][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:13:27,594][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:13:28,092][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:13:28,591][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:13:29,088][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:13:29,587][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:13:30,086][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:13:30,585][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:13:31,084][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:13:31,582][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:13:32,081][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:13:32,580][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:13:33,080][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:13:33,731][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 03:13:34,512][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:13:34,515][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:13:34,516][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:13:35,367][__main__][INFO] - Iteration 313 took 1m 13s (8.51% Gen, 90.33% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 57m 16s. Estimated total time: 61h 36m 10s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 12s, 500 more iterations: 10h 16m 1s. [2026-03-26 03:13:35,369][__main__][INFO] - Starting iteration 313. [2026-03-26 03:13:35,769][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:13:35,770][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:13:42,380][__main__][INFO] - Number of regex retries in iteration 313: 0 [2026-03-26 03:13:42,381][__main__][INFO] - agents played in iteration 313 are Alice, Bob [2026-03-26 03:13:43,253][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:13:43,298][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:13:43,298][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:13:43,299][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:13:43,905][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:13:44,363][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:13:44,867][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:13:45,367][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:13:45,867][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:13:46,367][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:13:46,867][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:13:47,366][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:13:47,866][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:13:48,367][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:13:48,865][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:13:49,364][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:13:49,861][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:13:50,360][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:13:50,857][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:13:51,356][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:13:51,855][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:13:52,355][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:13:52,854][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:13:53,353][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:13:53,851][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:13:54,350][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:13:54,850][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:13:55,350][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:13:55,849][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:13:56,348][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:13:56,846][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:13:57,344][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:13:57,843][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:13:58,342][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:13:58,840][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:13:59,340][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:13:59,839][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:14:00,339][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:14:00,837][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:14:01,335][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:14:01,832][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:14:02,329][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:14:02,826][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:14:03,324][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:14:03,824][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:14:04,322][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:14:04,821][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:14:05,319][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:14:05,817][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:14:06,314][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:14:06,812][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:14:07,310][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:14:07,807][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:14:08,306][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:14:08,805][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:14:09,303][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:14:09,802][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:14:10,300][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:14:10,798][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:14:11,298][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:14:11,797][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:14:12,296][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:14:12,795][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:14:13,294][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:14:13,793][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:14:14,291][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:14:14,790][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:14:15,289][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:14:15,788][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:14:16,287][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:14:16,787][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:14:17,286][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:14:17,786][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:14:18,287][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:14:18,787][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:14:19,287][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:14:19,786][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:14:20,285][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:14:20,783][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:14:21,281][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:14:21,778][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:14:22,276][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:14:22,772][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:14:23,270][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:14:23,767][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:14:24,267][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:14:24,766][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:14:25,266][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:14:25,765][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:14:26,264][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:14:26,765][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:14:27,265][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:14:27,765][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:14:28,263][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:14:28,762][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:14:29,260][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:14:29,758][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:14:30,255][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:14:30,752][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:14:31,249][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:14:31,748][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:14:32,247][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:14:32,747][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:14:33,246][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:14:33,745][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:14:34,244][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:14:34,743][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:14:35,242][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:14:35,741][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:14:36,239][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:14:36,738][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:14:37,237][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:14:37,735][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:14:38,232][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:14:38,730][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:14:39,228][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:14:39,727][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:14:40,226][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:14:40,724][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:14:41,223][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:14:41,722][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:14:42,221][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:14:42,720][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:14:43,220][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:14:43,718][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:14:44,220][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:14:44,720][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:14:45,220][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:14:45,723][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:14:46,225][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:14:46,725][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:14:47,225][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:14:47,725][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:14:48,357][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 03:14:49,263][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:14:49,266][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:14:49,268][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:14:50,013][__main__][INFO] - Iteration 314 took 1m 14s (8.90% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 12m 5s. Estimated total time: 61h 52m 13s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 44s, 500 more iterations: 10h 18m 42s. [2026-03-26 03:14:50,016][__main__][INFO] - Starting iteration 314. [2026-03-26 03:14:50,418][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:14:50,419][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:14:52,952][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:14:56,885][__main__][INFO] - Number of regex retries in iteration 314: 1 [2026-03-26 03:14:56,885][__main__][INFO] - agents played in iteration 314 are Alice, Bob [2026-03-26 03:14:57,758][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:14:57,802][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:14:57,803][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:14:57,803][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:14:58,410][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:14:58,865][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:14:59,370][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:14:59,870][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:15:00,369][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:15:00,868][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:15:01,369][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:15:01,868][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:15:02,367][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:15:02,868][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:15:03,367][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:15:03,867][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:15:04,366][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:15:04,866][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:15:05,365][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:15:05,865][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:15:06,365][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:15:06,865][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:15:07,364][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:15:07,864][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:15:08,364][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:15:08,863][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:15:09,363][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:15:09,862][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:15:10,361][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:15:10,863][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:15:11,364][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:15:11,865][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:15:12,365][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:15:12,866][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:15:13,366][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:15:13,867][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:15:14,367][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:15:14,867][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:15:15,365][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:15:15,864][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:15:16,363][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:15:16,862][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:15:17,360][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:15:17,859][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:15:18,361][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:15:18,860][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:15:19,360][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:15:19,860][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:15:20,359][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:15:20,858][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:15:21,357][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:15:21,856][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:15:22,355][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:15:22,854][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:15:23,351][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:15:23,849][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:15:24,347][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:15:24,845][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:15:25,344][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:15:25,843][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:15:26,341][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:15:26,841][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:15:27,340][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:15:27,840][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:15:28,339][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:15:28,838][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:15:29,337][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:15:29,836][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:15:30,335][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:15:30,833][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:15:31,332][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:15:31,831][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:15:32,330][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:15:32,828][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:15:33,327][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:15:33,826][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:15:34,325][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:15:34,825][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:15:35,325][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:15:35,824][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:15:36,324][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:15:36,823][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:15:37,320][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:15:37,820][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:15:38,320][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:15:38,819][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:15:39,319][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:15:39,818][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:15:40,317][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:15:40,816][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:15:41,315][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:15:41,814][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:15:42,313][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:15:42,812][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:15:43,313][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:15:43,812][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:15:44,312][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:15:44,811][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:15:45,310][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:15:45,810][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:15:46,310][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:15:46,809][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:15:47,308][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:15:47,807][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:15:48,308][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:15:48,806][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:15:49,305][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:15:49,805][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:15:50,304][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:15:50,803][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:15:51,302][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:15:51,802][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:15:52,301][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:15:52,800][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:15:53,300][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:15:53,799][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:15:54,299][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:15:54,798][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:15:55,298][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:15:55,796][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:15:56,293][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:15:56,791][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:15:57,289][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:15:57,787][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:15:58,286][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:15:58,785][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:15:59,284][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:15:59,783][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:16:00,282][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:16:00,781][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:16:01,281][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:16:01,780][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:16:02,280][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21734 tokens. [2026-03-26 03:16:02,919][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:16:03,719][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:16:03,721][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:16:03,722][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:16:04,450][__main__][INFO] - Iteration 315 took 1m 14s (8.73% Gen, 90.28% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 0m 15s. Estimated total time: 61h 41m 37s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 23s, 500 more iterations: 10h 16m 56s. [2026-03-26 03:16:04,452][__main__][INFO] - Starting iteration 315. [2026-03-26 03:16:04,853][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:16:04,854][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:16:11,372][__main__][INFO] - Number of regex retries in iteration 315: 0 [2026-03-26 03:16:11,372][__main__][INFO] - agents played in iteration 315 are Alice, Bob [2026-03-26 03:16:12,246][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:16:12,291][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:16:12,291][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:16:12,292][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:16:12,897][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:16:13,352][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:16:13,856][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:16:14,355][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:16:14,854][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:16:15,355][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:16:15,856][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:16:16,356][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:16:16,855][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:16:17,356][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:16:17,854][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:16:18,356][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:16:18,855][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:16:19,354][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:16:20,042][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:16:20,601][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:16:21,100][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:16:21,602][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:16:22,102][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:16:22,602][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:16:23,102][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:16:23,601][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:16:24,102][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:16:24,602][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:16:25,103][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:16:25,601][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:16:26,100][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:16:26,598][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:16:27,095][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:16:27,593][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:16:28,092][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:16:28,590][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:16:29,088][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:16:29,586][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:16:30,084][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:16:30,581][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:16:31,079][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:16:31,576][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:16:32,074][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:16:32,571][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:16:33,068][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:16:33,568][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:16:34,068][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:16:34,567][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:16:35,066][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:16:35,566][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:16:36,066][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:16:36,566][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:16:37,066][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:16:37,565][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:16:38,064][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:16:38,562][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:16:39,060][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:16:39,557][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:16:40,056][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:16:40,555][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:16:41,053][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:16:41,556][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:16:42,058][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:16:42,560][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:16:43,061][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:16:43,563][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:16:44,064][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:16:44,566][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:16:45,068][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:16:45,567][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:16:46,067][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:16:46,566][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:16:47,065][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:16:47,565][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:16:48,065][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:16:48,567][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:16:49,066][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:16:49,566][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:16:50,066][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:16:50,566][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:16:51,065][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:16:51,565][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:16:52,065][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:16:52,564][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:16:53,063][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:16:53,564][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:16:54,065][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:16:54,565][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:16:55,067][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:16:55,567][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:16:56,067][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:16:56,567][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:16:57,068][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:16:57,566][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:16:58,064][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:16:58,563][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:16:59,061][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:16:59,559][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:17:00,058][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:17:00,555][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:17:01,053][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:17:01,551][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:17:02,048][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:17:02,546][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:17:03,044][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:17:03,542][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:17:04,042][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:17:04,542][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:17:05,041][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:17:05,541][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:17:06,041][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:17:06,541][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:17:07,041][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:17:07,540][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:17:08,040][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:17:08,540][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:17:09,042][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:17:09,541][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:17:10,039][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:17:10,538][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:17:11,036][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:17:11,534][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:17:12,031][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:17:12,529][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:17:13,027][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:17:13,529][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:17:14,032][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:17:14,533][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:17:15,035][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:17:15,537][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:17:16,041][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:17:16,543][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:17:17,047][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 03:17:17,705][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 03:17:18,491][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:17:18,493][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:17:18,495][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:17:19,178][__main__][INFO] - Iteration 316 took 1m 14s (8.77% Gen, 90.31% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 13m 40s. Estimated total time: 61h 56m 17s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 52s, 500 more iterations: 10h 19m 22s. [2026-03-26 03:17:19,187][__main__][INFO] - Starting iteration 316. [2026-03-26 03:17:19,588][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:17:19,588][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:17:21,825][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:17:26,156][__main__][INFO] - Number of regex retries in iteration 316: 1 [2026-03-26 03:17:26,156][__main__][INFO] - agents played in iteration 316 are Alice, Bob [2026-03-26 03:17:27,167][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:17:27,211][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:17:27,212][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:17:27,212][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:17:27,858][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:17:28,313][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:17:28,816][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:17:29,315][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:17:29,814][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:17:30,313][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:17:30,812][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:17:31,311][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:17:31,810][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:17:32,309][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:17:32,808][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:17:33,306][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:17:33,805][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:17:34,305][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:17:34,805][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:17:35,304][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:17:35,803][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:17:36,302][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:17:36,802][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:17:37,302][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:17:37,801][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:17:38,302][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:17:38,801][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:17:39,300][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:17:39,800][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:17:40,298][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:17:40,796][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:17:41,293][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:17:41,792][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:17:42,290][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:17:42,788][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:17:43,287][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:17:43,786][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:17:44,286][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:17:44,785][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:17:45,285][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:17:45,787][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:17:46,286][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:17:46,785][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:17:47,284][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:17:47,783][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:17:48,286][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:17:48,787][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:17:49,286][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:17:49,787][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:17:50,287][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:17:50,787][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:17:51,286][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:17:51,786][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:17:52,285][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:17:52,783][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:17:53,282][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:17:53,781][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:17:54,279][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:17:54,780][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:17:55,279][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:17:55,778][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:17:56,277][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:17:56,776][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:17:57,276][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:17:57,775][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:17:58,275][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:17:58,775][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:17:59,274][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:17:59,773][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:18:00,272][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:18:00,771][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:18:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:18:01,769][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:18:02,268][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:18:02,768][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:18:03,266][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:18:03,766][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:18:04,266][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:18:04,764][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:18:05,262][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:18:05,761][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:18:06,260][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:18:06,758][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:18:07,255][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:18:07,753][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:18:08,252][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:18:08,751][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:18:09,250][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:18:09,750][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:18:10,249][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:18:10,748][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:18:11,248][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:18:11,748][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:18:12,247][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:18:12,747][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:18:13,245][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:18:13,744][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:18:14,242][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:18:14,741][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:18:15,240][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:18:15,741][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:18:16,240][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:18:16,739][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:18:17,239][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:18:17,739][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:18:18,240][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:18:18,740][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:18:19,240][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:18:19,741][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:18:20,241][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:18:20,742][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:18:21,242][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:18:21,742][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:18:22,244][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:18:22,744][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:18:23,244][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:18:23,744][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:18:24,242][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:18:24,741][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:18:25,241][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:18:25,740][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:18:26,241][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:18:26,740][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:18:27,241][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:18:27,741][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:18:28,241][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:18:28,742][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:18:29,242][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:18:29,742][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:18:30,243][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:18:30,743][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:18:31,243][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:18:31,743][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:18:32,375][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:18:33,120][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:18:33,123][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:18:33,124][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:18:33,785][__main__][INFO] - Iteration 317 took 1m 14s (8.85% Gen, 90.26% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 6m 1s. Estimated total time: 61h 49m 53s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 18s. [2026-03-26 03:18:33,787][__main__][INFO] - Starting iteration 317. [2026-03-26 03:18:34,193][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:18:34,194][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:18:36,978][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:18:38,626][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:18:39,840][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:18:40,632][__main__][INFO] - Number of regex retries in iteration 317: 3 [2026-03-26 03:18:40,633][__main__][INFO] - agents played in iteration 317 are Alice, Bob [2026-03-26 03:18:41,507][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:18:41,831][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:18:41,832][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:18:41,833][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:18:42,439][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:18:42,892][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:18:43,394][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:18:43,892][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:18:44,389][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:18:44,887][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:18:45,386][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:18:45,883][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:18:46,380][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:18:46,881][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:18:47,381][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:18:47,881][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:18:48,383][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:18:48,882][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:18:49,381][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:18:49,880][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:18:50,378][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:18:50,878][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:18:51,381][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:18:51,882][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:18:52,382][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:18:52,882][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:18:53,382][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:18:53,883][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:18:54,383][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:18:54,882][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:18:55,383][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:18:55,881][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:18:56,382][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:18:56,880][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:18:57,379][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:18:57,877][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:18:58,377][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:18:58,877][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:18:59,378][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:18:59,879][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:19:00,379][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:19:00,879][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:19:01,378][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:19:01,879][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:19:02,380][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:19:02,879][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:19:03,378][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:19:03,877][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:19:04,375][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:19:04,873][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:19:05,372][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:19:05,872][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:19:06,371][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:19:06,871][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:19:07,369][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:19:07,868][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:19:08,367][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:19:08,866][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:19:09,365][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:19:09,864][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:19:10,363][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:19:10,862][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:19:11,361][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:19:11,860][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:19:12,359][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:19:12,857][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:19:13,355][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:19:13,855][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:19:14,354][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:19:14,852][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:19:15,351][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:19:15,852][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:19:16,351][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:19:16,849][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:19:17,347][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:19:17,846][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:19:18,347][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:19:18,846][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:19:19,346][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:19:19,845][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:19:20,344][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:19:20,843][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:19:21,343][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:19:21,843][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:19:22,342][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:19:22,842][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:19:23,343][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:19:23,843][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:19:24,344][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:19:24,845][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:19:25,345][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:19:25,846][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:19:26,346][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:19:26,845][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:19:27,345][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:19:27,843][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:19:28,344][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:19:28,844][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:19:29,343][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:19:29,842][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:19:30,341][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:19:30,842][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:19:31,343][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:19:31,843][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:19:32,343][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:19:32,842][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:19:33,342][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:19:33,844][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:19:34,344][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:19:34,842][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:19:35,342][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:19:35,842][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:19:36,342][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:19:36,841][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:19:37,339][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:19:37,838][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:19:38,337][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:19:38,836][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:19:39,334][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:19:39,833][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:19:40,332][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:19:40,830][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:19:41,329][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:19:41,827][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:19:42,326][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:19:42,825][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:19:43,324][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:19:43,823][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:19:44,322][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:19:44,821][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:19:45,319][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:19:45,819][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:19:46,318][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:19:46,953][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 03:19:47,703][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:19:47,706][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:19:47,707][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:19:48,384][__main__][INFO] - Iteration 318 took 1m 14s (8.68% Gen, 90.41% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 4m 28s. Estimated total time: 61h 49m 34s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 15s. [2026-03-26 03:19:48,386][__main__][INFO] - Starting iteration 318. [2026-03-26 03:19:48,788][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:19:48,788][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:19:50,278][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:19:55,020][__main__][INFO] - Number of regex retries in iteration 318: 1 [2026-03-26 03:19:55,021][__main__][INFO] - agents played in iteration 318 are Alice, Bob [2026-03-26 03:19:55,894][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:19:55,938][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:19:55,939][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:19:55,939][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:19:56,549][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:19:57,003][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:19:57,507][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:19:58,006][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:19:58,504][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:19:59,003][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:19:59,502][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:20:00,000][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:20:00,498][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:20:00,997][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:20:01,496][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:20:01,994][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:20:02,492][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:20:02,989][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:20:03,488][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:20:03,987][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:20:04,486][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:20:04,985][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:20:05,484][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:20:05,983][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:20:06,485][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:20:06,985][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:20:07,485][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:20:07,985][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:20:08,484][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:20:08,983][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:20:09,482][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:20:09,980][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:20:10,481][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:20:10,980][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:20:11,479][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:20:11,979][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:20:12,478][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:20:12,977][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:20:13,476][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:20:13,975][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:20:14,473][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:20:14,972][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:20:15,472][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:20:15,971][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:20:16,470][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:20:16,969][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:20:17,468][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:20:17,968][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:20:18,469][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:20:18,968][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:20:19,467][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:20:19,967][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:20:20,467][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:20:20,966][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:20:21,466][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:20:21,967][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:20:22,467][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:20:22,966][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:20:23,466][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:20:23,965][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:20:24,465][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:20:24,966][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:20:25,468][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:20:25,968][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:20:26,467][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:20:26,967][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:20:27,468][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:20:27,968][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:20:28,468][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:20:28,967][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:20:29,465][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:20:29,964][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:20:30,463][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:20:30,961][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:20:31,460][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:20:31,957][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:20:32,456][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:20:32,954][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:20:33,453][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:20:33,954][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:20:34,453][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:20:34,952][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:20:35,451][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:20:35,950][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:20:36,450][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:20:36,950][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:20:37,450][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:20:37,950][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:20:38,449][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:20:38,948][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:20:39,447][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:20:39,948][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:20:40,448][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:20:40,947][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:20:41,445][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:20:41,944][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:20:42,444][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:20:42,944][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:20:43,443][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:20:43,944][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:20:44,443][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:20:44,942][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:20:45,444][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:20:45,947][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:20:46,448][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:20:46,949][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:20:47,449][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:20:47,950][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:20:48,451][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:20:48,950][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:20:49,449][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:20:49,948][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:20:50,447][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:20:50,945][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:20:51,444][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:20:51,943][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:20:52,443][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:20:52,942][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:20:53,442][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:20:53,944][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:20:54,443][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:20:54,942][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:20:55,442][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:20:55,943][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:20:56,442][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:20:56,942][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:20:57,442][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:20:57,944][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:20:58,443][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:20:58,943][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:20:59,443][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:20:59,942][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:21:00,443][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 03:21:01,079][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 03:21:01,814][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:21:01,816][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:21:01,818][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:21:02,469][__main__][INFO] - Iteration 319 took 1m 13s (8.46% Gen, 90.66% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 37m 45s. Estimated total time: 61h 24m 5s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 48s, 500 more iterations: 10h 14m 0s. [2026-03-26 03:21:02,471][__main__][INFO] - Starting iteration 319. [2026-03-26 03:21:02,873][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:21:02,874][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:21:09,284][__main__][INFO] - Number of regex retries in iteration 319: 0 [2026-03-26 03:21:09,285][__main__][INFO] - agents played in iteration 319 are Alice, Bob [2026-03-26 03:21:10,164][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:21:10,208][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:21:10,209][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:21:10,210][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:21:10,820][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:21:11,274][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:21:11,777][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:21:12,275][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:21:12,773][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:21:13,272][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:21:13,770][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:21:14,269][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:21:14,768][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:21:15,269][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:21:15,769][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:21:16,269][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:21:16,769][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:21:17,268][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:21:17,768][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:21:18,270][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:21:18,771][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:21:19,270][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:21:19,770][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:21:20,270][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:21:20,769][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:21:21,268][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:21:21,768][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:21:22,268][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:21:22,768][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:21:23,267][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:21:23,767][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:21:24,268][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:21:24,767][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:21:25,268][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:21:25,767][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:21:26,267][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:21:26,768][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:21:27,267][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:21:27,767][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:21:28,267][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:21:28,767][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:21:29,266][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:21:29,765][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:21:30,264][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:21:30,763][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:21:31,264][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:21:31,764][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:21:32,264][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:21:32,764][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:21:33,264][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:21:33,763][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:21:34,263][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:21:34,763][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:21:35,262][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:21:35,762][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:21:36,261][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:21:36,760][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:21:37,259][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:21:37,758][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:21:38,257][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:21:38,756][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:21:39,255][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:21:39,754][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:21:40,254][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:21:40,755][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:21:41,254][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:21:41,753][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:21:42,253][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:21:42,753][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:21:43,252][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:21:43,750][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:21:44,249][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:21:44,748][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:21:45,247][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:21:45,745][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:21:46,245][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:21:46,745][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:21:47,245][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:21:47,743][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:21:48,246][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:21:48,747][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:21:49,247][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:21:49,748][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:21:50,249][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:21:50,750][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:21:51,249][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:21:51,749][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:21:52,249][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:21:52,748][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:21:53,248][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:21:53,749][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:21:54,249][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:21:54,749][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:21:55,249][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:21:55,749][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:21:56,249][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:21:56,749][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:21:57,248][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:21:57,748][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:21:58,248][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:21:58,747][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:21:59,246][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:21:59,745][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:22:00,244][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:22:00,743][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:22:01,243][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:22:01,743][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:22:02,242][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:22:02,740][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:22:03,240][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:22:03,742][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:22:04,243][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:22:04,743][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:22:05,243][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:22:05,744][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:22:06,245][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:22:06,745][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:22:07,244][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:22:07,743][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:22:08,242][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:22:08,743][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:22:09,242][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:22:09,741][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:22:10,240][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:22:10,739][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:22:11,239][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:22:11,739][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:22:12,238][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:22:12,737][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:22:13,236][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:22:13,735][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:22:14,234][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:22:14,733][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 03:22:15,371][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:22:16,116][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:22:16,118][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:22:16,120][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:22:16,769][__main__][INFO] - Iteration 320 took 1m 13s (8.68% Gen, 90.44% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 47m 15s. Estimated total time: 61h 34m 50s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 9s, 500 more iterations: 10h 15m 48s. [2026-03-26 03:22:16,771][__main__][INFO] - Starting iteration 320. [2026-03-26 03:22:17,170][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:22:17,170][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:22:21,100][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:22:23,908][__main__][INFO] - Number of regex retries in iteration 320: 1 [2026-03-26 03:22:23,909][__main__][INFO] - agents played in iteration 320 are Alice, Bob [2026-03-26 03:22:24,782][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:22:24,827][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:22:24,827][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:22:24,828][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:22:25,430][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:22:25,886][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:22:26,388][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:22:26,888][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:22:27,387][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:22:27,887][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:22:28,386][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:22:28,885][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:22:29,384][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:22:29,883][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:22:30,382][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:22:30,881][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:22:31,379][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:22:31,877][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:22:32,376][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:22:32,875][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:22:33,374][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:22:33,875][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:22:34,375][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:22:34,873][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:22:35,373][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:22:35,872][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:22:36,372][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:22:36,871][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:22:37,371][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:22:37,868][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:22:38,366][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:22:38,863][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:22:39,360][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:22:39,858][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:22:40,357][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:22:40,854][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:22:41,353][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:22:41,851][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:22:42,350][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:22:42,848][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:22:43,346][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:22:43,845][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:22:44,343][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:22:44,842][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:22:45,340][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:22:45,840][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:22:46,339][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:22:46,837][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:22:47,336][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:22:47,835][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:22:48,336][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:22:48,835][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:22:49,335][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:22:49,833][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:22:50,333][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:22:50,831][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:22:51,330][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:22:51,830][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:22:52,330][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:22:52,830][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:22:53,329][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:22:53,828][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:22:54,327][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:22:54,826][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:22:55,326][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:22:55,826][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:22:56,326][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:22:56,825][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:22:57,324][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:22:57,822][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:22:58,321][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:22:58,820][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:22:59,321][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:22:59,821][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:23:00,320][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:23:00,819][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:23:01,317][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:23:01,815][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:23:02,313][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:23:02,810][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:23:03,309][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:23:03,807][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:23:04,306][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:23:04,804][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:23:05,302][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:23:05,801][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:23:06,300][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:23:06,799][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:23:07,299][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:23:07,799][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:23:08,299][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:23:08,798][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:23:09,297][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:23:09,797][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:23:10,294][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:23:10,793][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:23:11,291][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:23:11,790][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:23:12,288][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:23:12,786][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:23:13,284][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:23:13,782][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:23:14,279][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:23:14,777][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:23:15,275][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:23:15,773][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:23:16,271][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:23:16,770][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:23:17,268][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:23:17,768][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:23:18,269][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:23:18,769][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:23:19,269][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:23:19,768][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:23:20,267][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:23:20,766][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:23:21,266][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:23:21,765][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:23:22,265][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:23:22,763][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:23:23,264][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:23:23,764][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:23:24,263][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:23:24,762][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:23:25,262][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:23:25,760][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:23:26,259][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:23:26,758][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:23:27,257][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:23:27,756][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:23:28,256][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:23:28,754][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:23:29,253][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:23:29,886][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:23:30,658][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:23:30,660][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:23:30,661][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:23:31,803][__main__][INFO] - Iteration 321 took 1m 14s (9.03% Gen, 89.44% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 22m 53s. Estimated total time: 62h 11m 42s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 23s, 500 more iterations: 10h 21m 57s. [2026-03-26 03:23:31,806][__main__][INFO] - Starting iteration 321. [2026-03-26 03:23:32,204][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:23:32,205][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:23:32,798][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:23:38,064][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:23:38,824][__main__][INFO] - Number of regex retries in iteration 321: 2 [2026-03-26 03:23:38,825][__main__][INFO] - agents played in iteration 321 are Alice, Bob [2026-03-26 03:23:39,714][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:23:39,758][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:23:39,759][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:23:39,759][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:23:40,367][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:23:40,818][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:23:41,321][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:23:41,818][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:23:42,315][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:23:42,812][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:23:43,309][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:23:43,807][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:23:44,304][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:23:44,802][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:23:45,301][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:23:45,803][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:23:46,303][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:23:46,802][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:23:47,300][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:23:47,799][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:23:48,300][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:23:48,799][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:23:49,300][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:23:49,799][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:23:50,299][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:23:50,797][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:23:51,295][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:23:51,794][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:23:52,292][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:23:52,791][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:23:53,291][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:23:53,789][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:23:54,289][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:23:54,787][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:23:55,286][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:23:55,785][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:23:56,284][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:23:56,779][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:23:57,276][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:23:57,773][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:23:58,270][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:23:58,767][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:23:59,265][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:23:59,761][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:24:00,259][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:24:00,757][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:24:01,255][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:24:01,754][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:24:02,253][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:24:02,751][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:24:03,250][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:24:03,749][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:24:04,248][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:24:04,748][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:24:05,249][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:24:05,750][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:24:06,251][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:24:06,751][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:24:07,250][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:24:07,751][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:24:08,252][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:24:08,751][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:24:09,252][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:24:09,751][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:24:10,251][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:24:10,751][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:24:11,252][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:24:11,752][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:24:12,252][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:24:12,750][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:24:13,247][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:24:13,745][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:24:14,243][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:24:14,740][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:24:15,236][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:24:15,732][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:24:16,229][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:24:16,729][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:24:17,226][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:24:17,725][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:24:18,227][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:24:18,726][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:24:19,223][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:24:19,720][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:24:20,218][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:24:20,716][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:24:21,214][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:24:21,713][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:24:22,211][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:24:22,710][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:24:23,209][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:24:23,707][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:24:24,206][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:24:24,706][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:24:25,206][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:24:25,707][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:24:26,206][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:24:26,706][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:24:27,205][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:24:27,705][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:24:28,205][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:24:28,700][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:24:29,196][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:24:29,693][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:24:30,190][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:24:30,687][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:24:31,183][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:24:31,681][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:24:32,178][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:24:32,676][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:24:33,174][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:24:33,673][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:24:34,172][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:24:34,671][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:24:35,169][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:24:35,668][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:24:36,167][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:24:36,666][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:24:37,165][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:24:37,664][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:24:38,165][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:24:38,665][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:24:39,164][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:24:39,664][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:24:40,163][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:24:40,662][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:24:41,161][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:24:41,662][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:24:42,162][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:24:42,662][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:24:43,161][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:24:43,661][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:24:44,160][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:24:44,795][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 03:24:45,725][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:24:45,730][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:24:45,733][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:24:46,492][__main__][INFO] - Iteration 322 took 1m 14s (8.91% Gen, 90.06% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 4m 22s. Estimated total time: 61h 54m 26s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 48s, 500 more iterations: 10h 19m 4s. [2026-03-26 03:24:46,494][__main__][INFO] - Starting iteration 322. [2026-03-26 03:24:46,895][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:24:46,896][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:24:53,287][__main__][INFO] - Number of regex retries in iteration 322: 0 [2026-03-26 03:24:53,288][__main__][INFO] - agents played in iteration 322 are Alice, Bob [2026-03-26 03:24:54,171][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:24:54,215][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:24:54,216][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:24:54,216][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:24:54,823][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:24:55,278][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:24:55,783][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:24:56,283][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:24:56,784][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:24:57,285][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:24:57,784][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:24:58,284][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:24:58,783][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:24:59,281][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:24:59,779][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:25:00,277][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:25:00,776][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:25:01,274][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:25:01,773][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:25:02,272][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:25:02,770][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:25:03,267][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:25:03,764][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:25:04,262][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:25:04,760][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:25:05,258][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:25:05,755][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:25:06,253][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:25:06,751][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:25:07,251][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:25:07,750][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:25:08,249][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:25:08,749][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:25:09,249][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:25:09,749][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:25:10,249][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:25:10,747][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:25:11,246][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:25:11,746][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:25:12,246][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:25:12,745][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:25:13,244][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:25:13,743][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:25:14,242][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:25:14,741][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:25:15,238][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:25:15,739][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:25:16,239][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:25:16,737][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:25:17,235][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:25:17,733][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:25:18,232][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:25:18,731][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:25:19,229][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:25:19,727][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:25:20,225][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:25:20,723][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:25:21,222][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:25:21,720][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:25:22,217][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:25:22,713][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:25:23,211][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:25:23,709][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:25:24,206][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:25:24,705][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:25:25,204][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:25:25,702][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:25:26,200][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:25:26,698][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:25:27,196][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:25:27,695][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:25:28,193][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:25:28,691][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:25:29,190][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:25:29,689][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:25:30,188][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:25:30,687][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:25:31,185][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:25:31,683][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:25:32,181][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:25:32,679][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:25:33,176][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:25:33,674][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:25:34,172][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:25:34,671][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:25:35,168][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:25:35,666][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:25:36,163][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:25:36,660][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:25:37,157][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:25:37,654][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:25:38,152][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:25:38,650][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:25:39,149][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:25:39,648][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:25:40,147][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:25:40,646][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:25:41,146][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:25:41,645][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:25:42,145][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:25:42,644][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:25:43,143][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:25:43,643][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:25:44,141][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:25:44,639][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:25:45,137][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:25:45,635][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:25:46,133][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:25:46,632][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:25:47,131][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:25:47,629][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:25:48,128][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:25:48,628][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:25:49,127][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:25:49,625][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:25:50,124][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:25:50,624][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:25:51,122][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:25:51,620][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:25:52,117][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:25:52,614][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:25:53,111][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:25:53,608][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:25:54,106][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:25:54,605][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:25:55,105][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:25:55,603][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:25:56,102][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:25:56,600][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:25:57,098][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:25:57,596][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:25:58,094][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:25:58,592][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 03:25:59,220][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 03:26:00,103][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:26:00,105][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:26:00,107][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:26:00,757][__main__][INFO] - Iteration 323 took 1m 13s (8.65% Gen, 90.46% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 41m 48s. Estimated total time: 61h 33m 7s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 6s, 500 more iterations: 10h 15m 31s. [2026-03-26 03:26:00,759][__main__][INFO] - Starting iteration 323. [2026-03-26 03:26:01,157][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:26:01,158][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:26:01,738][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:26:08,168][__main__][INFO] - Number of regex retries in iteration 323: 1 [2026-03-26 03:26:08,169][__main__][INFO] - agents played in iteration 323 are Alice, Bob [2026-03-26 03:26:09,047][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:26:09,092][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:26:09,092][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:26:09,093][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:26:09,700][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:26:10,153][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:26:10,653][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:26:11,148][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:26:11,643][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:26:12,137][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:26:12,633][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:26:13,128][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:26:13,626][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:26:14,126][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:26:14,628][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:26:15,129][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:26:15,630][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:26:16,130][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:26:16,630][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:26:17,130][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:26:17,630][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:26:18,131][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:26:18,632][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:26:19,132][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:26:19,631][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:26:20,130][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:26:20,629][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:26:21,129][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:26:21,630][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:26:22,129][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:26:22,629][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:26:23,128][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:26:23,627][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:26:24,128][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:26:24,627][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:26:25,128][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:26:25,628][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:26:26,127][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:26:26,626][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:26:27,126][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:26:27,626][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:26:28,126][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:26:28,625][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:26:29,125][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:26:29,625][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:26:30,123][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:26:30,622][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:26:31,120][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:26:31,617][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:26:32,115][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:26:32,613][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:26:33,111][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:26:33,610][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:26:34,109][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:26:34,608][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:26:35,107][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:26:35,605][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:26:36,105][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:26:36,605][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:26:37,105][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:26:37,604][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:26:38,104][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:26:38,602][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:26:39,100][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:26:39,598][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:26:40,097][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:26:40,595][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:26:41,094][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:26:41,592][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:26:42,086][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:26:42,580][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:26:43,074][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:26:43,569][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:26:44,063][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:26:44,556][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:26:45,051][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:26:45,548][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:26:46,049][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:26:46,551][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:26:47,051][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:26:47,552][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:26:48,052][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:26:48,553][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:26:49,054][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:26:49,555][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:26:50,055][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:26:50,555][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:26:51,056][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:26:51,556][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:26:52,056][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:26:52,556][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:26:53,057][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:26:53,556][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:26:54,056][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:26:54,556][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:26:55,056][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:26:55,556][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:26:56,054][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:26:56,554][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:26:57,054][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:26:57,554][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:26:58,053][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:26:58,552][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:26:59,052][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:26:59,551][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:27:00,051][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:27:00,551][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:27:01,052][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:27:01,552][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:27:02,050][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:27:02,549][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:27:03,049][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:27:03,547][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:27:04,046][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:27:04,545][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:27:05,044][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:27:05,543][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:27:06,041][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:27:06,540][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:27:07,038][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:27:07,537][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:27:08,035][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:27:08,533][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:27:09,033][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:27:09,531][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:27:10,030][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:27:10,528][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:27:11,028][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:27:11,526][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:27:12,026][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:27:12,524][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:27:13,023][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:27:13,523][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 03:27:14,156][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.39%, ΔTime: 00:01:04 [2026-03-26 03:27:15,124][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:27:15,127][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:27:15,128][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:27:15,777][__main__][INFO] - Iteration 324 took 1m 14s (9.40% Gen, 89.73% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 55h 18m 27s. Estimated total time: 62h 11m 1s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 22s, 500 more iterations: 10h 21m 50s. [2026-03-26 03:27:15,779][__main__][INFO] - Starting iteration 324. [2026-03-26 03:27:16,310][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:27:16,311][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:27:23,295][__main__][INFO] - Number of regex retries in iteration 324: 0 [2026-03-26 03:27:23,295][__main__][INFO] - agents played in iteration 324 are Alice, Bob [2026-03-26 03:27:24,168][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:27:24,213][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:27:24,213][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:27:24,214][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:27:24,816][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:27:25,271][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:27:25,774][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:27:26,273][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:27:26,772][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:27:27,271][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:27:27,770][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:27:28,269][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:27:28,768][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:27:29,267][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:27:29,766][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:27:30,264][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:27:30,762][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:27:31,261][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:27:31,760][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:27:32,259][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:27:32,757][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:27:33,256][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:27:33,754][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:27:34,252][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:27:34,750][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:27:35,249][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:27:35,747][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:27:36,246][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:27:36,745][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:27:37,246][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:27:37,745][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:27:38,243][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:27:38,742][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:27:39,240][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:27:39,739][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:27:40,238][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:27:40,736][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:27:41,235][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:27:41,732][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:27:42,229][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:27:42,727][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:27:43,226][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:27:43,723][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:27:44,223][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:27:44,722][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:27:45,221][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:27:45,719][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:27:46,217][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:27:46,715][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:27:47,213][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:27:47,711][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:27:48,212][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:27:48,711][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:27:49,210][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:27:49,709][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:27:50,207][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:27:50,706][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:27:51,205][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:27:51,703][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:27:52,203][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:27:52,702][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:27:53,199][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:27:53,697][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:27:54,194][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:27:54,691][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:27:55,189][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:27:55,688][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:27:56,184][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:27:56,684][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:27:57,183][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:27:57,683][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:27:58,183][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:27:58,681][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:27:59,179][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:27:59,678][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:28:00,176][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:28:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:28:01,173][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:28:01,672][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:28:02,171][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:28:02,666][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:28:03,166][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:28:03,665][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:28:04,164][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:28:04,664][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:28:05,163][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:28:05,661][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:28:06,159][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:28:06,657][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:28:07,155][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:28:07,653][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:28:08,152][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:28:08,650][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:28:09,150][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:28:09,648][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:28:10,147][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:28:10,645][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:28:11,144][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:28:11,643][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:28:12,142][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:28:12,642][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:28:13,140][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:28:13,638][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:28:14,136][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:28:14,633][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:28:15,131][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:28:15,632][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:28:16,132][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:28:16,630][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:28:17,130][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:28:17,628][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:28:18,128][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:28:18,629][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:28:19,127][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:28:19,626][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:28:20,125][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:28:20,626][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:28:21,126][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:28:21,625][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:28:22,125][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:28:22,623][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:28:23,122][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:28:23,621][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:28:24,119][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:28:24,619][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:28:25,117][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:28:25,615][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:28:26,113][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:28:26,611][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:28:27,109][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:28:27,606][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:28:28,104][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:28:28,604][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21722 tokens. [2026-03-26 03:28:29,229][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.16%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 03:28:30,003][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:28:30,005][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:28:30,007][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:28:30,887][__main__][INFO] - Iteration 325 took 1m 14s (9.35% Gen, 89.30% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 21m 37s. Estimated total time: 62h 15m 25s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 30s, 500 more iterations: 10h 22m 34s. [2026-03-26 03:28:30,889][__main__][INFO] - Starting iteration 325. [2026-03-26 03:28:31,287][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:28:31,288][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:28:31,871][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:28:38,395][__main__][INFO] - Number of regex retries in iteration 325: 1 [2026-03-26 03:28:38,396][__main__][INFO] - agents played in iteration 325 are Alice, Bob [2026-03-26 03:28:39,272][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:28:39,316][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:28:39,317][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:28:39,317][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:28:39,921][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:28:40,375][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:28:40,879][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:28:41,377][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:28:41,875][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:28:42,374][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:28:42,872][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:28:43,371][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:28:43,870][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:28:44,371][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:28:44,871][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:28:45,370][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:28:45,869][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:28:46,369][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:28:46,868][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:28:47,367][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:28:47,868][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:28:48,369][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:28:48,869][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:28:49,367][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:28:49,866][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:28:50,365][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:28:50,865][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:28:51,365][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:28:51,865][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:28:52,364][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:28:52,863][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:28:53,362][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:28:53,860][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:28:54,359][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:28:54,859][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:28:55,358][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:28:55,857][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:28:56,357][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:28:56,855][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:28:57,354][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:28:57,852][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:28:58,352][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:28:58,852][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:28:59,352][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:28:59,852][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:29:00,353][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:29:00,852][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:29:01,352][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:29:01,852][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:29:02,352][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:29:02,851][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:29:03,351][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:29:03,850][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:29:04,351][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:29:04,850][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:29:05,350][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:29:05,851][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:29:06,352][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:29:06,852][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:29:07,351][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:29:07,851][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:29:08,351][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:29:08,850][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:29:09,350][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:29:09,849][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:29:10,349][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:29:10,848][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:29:11,347][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:29:11,847][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:29:12,345][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:29:12,844][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:29:13,342][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:29:13,840][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:29:14,338][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:29:14,835][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:29:15,333][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:29:15,831][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:29:16,330][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:29:16,830][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:29:17,331][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:29:17,832][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:29:18,331][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:29:18,831][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:29:19,330][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:29:19,830][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:29:20,329][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:29:20,828][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:29:21,327][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:29:21,825][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:29:22,324][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:29:22,822][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:29:23,320][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:29:23,817][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:29:24,314][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:29:24,813][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:29:25,313][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:29:25,812][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:29:26,310][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:29:26,809][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:29:27,308][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:29:27,807][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:29:28,307][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:29:28,806][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:29:29,306][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:29:29,805][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:29:30,304][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:29:30,803][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:29:31,302][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:29:31,800][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:29:32,300][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:29:32,799][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:29:33,299][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:29:33,799][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:29:34,298][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:29:34,798][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:29:35,297][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:29:35,796][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:29:36,295][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:29:36,794][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:29:37,294][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:29:37,794][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:29:38,293][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:29:38,792][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:29:39,291][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:29:39,791][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:29:40,290][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:29:40,790][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:29:41,290][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:29:41,788][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:29:42,287][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:29:42,787][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:29:43,286][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:29:43,786][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:29:44,417][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 03:29:45,163][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:29:45,166][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:29:45,167][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:29:45,821][__main__][INFO] - Iteration 326 took 1m 14s (9.54% Gen, 89.58% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 55h 11m 41s. Estimated total time: 62h 6m 45s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 13s, 500 more iterations: 10h 21m 7s. [2026-03-26 03:29:45,823][__main__][INFO] - Starting iteration 326. [2026-03-26 03:29:46,227][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:29:46,228][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:29:51,548][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:29:53,311][__main__][INFO] - Number of regex retries in iteration 326: 1 [2026-03-26 03:29:53,312][__main__][INFO] - agents played in iteration 326 are Alice, Bob [2026-03-26 03:29:54,197][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:29:54,242][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:29:54,242][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:29:54,243][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:29:54,857][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:29:55,314][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:29:55,816][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:29:56,314][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:29:56,813][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:29:57,313][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:29:57,811][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:29:58,310][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:29:58,809][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:29:59,307][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:29:59,805][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:30:00,305][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:30:00,804][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:30:01,303][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:30:01,800][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:30:02,298][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:30:02,796][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:30:03,295][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:30:03,794][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:30:04,294][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:30:04,794][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:30:05,294][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:30:05,794][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:30:06,292][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:30:06,791][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:30:07,290][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:30:07,789][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:30:08,287][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:30:08,785][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:30:09,284][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:30:09,783][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:30:10,281][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:30:10,779][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:30:11,277][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:30:11,775][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:30:12,272][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:30:12,771][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:30:13,269][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:30:13,766][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:30:14,264][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:30:14,761][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:30:15,259][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:30:15,756][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:30:16,254][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:30:16,753][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:30:17,252][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:30:17,751][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:30:18,252][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:30:18,753][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:30:19,255][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:30:19,754][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:30:20,255][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:30:20,754][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:30:21,255][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:30:21,754][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:30:22,253][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:30:22,752][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:30:23,256][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:30:23,758][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:30:24,261][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:30:24,763][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:30:25,264][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:30:25,765][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:30:26,267][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:30:26,769][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:30:27,269][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:30:27,769][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:30:28,268][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:30:28,767][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:30:29,266][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:30:29,766][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:30:30,266][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:30:30,766][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:30:31,264][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:30:31,762][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:30:32,259][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:30:32,756][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:30:33,254][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:30:33,751][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:30:34,248][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:30:34,747][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:30:35,247][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:30:35,746][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:30:36,245][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:30:36,744][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:30:37,243][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:30:37,743][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:30:38,242][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:30:38,741][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:30:39,239][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:30:39,737][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:30:40,232][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:30:40,730][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:30:41,228][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:30:41,727][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:30:42,225][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:30:42,726][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:30:43,224][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:30:43,722][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:30:44,219][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:30:44,716][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:30:45,214][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:30:45,711][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:30:46,207][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:30:46,705][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:30:47,205][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:30:47,704][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:30:48,204][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:30:48,702][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:30:49,199][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:30:49,697][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:30:50,194][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:30:50,691][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:30:51,190][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:30:51,689][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:30:52,188][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:30:52,686][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:30:53,185][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:30:53,684][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:30:54,182][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:30:54,679][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:30:55,181][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:30:55,684][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:30:56,186][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:30:56,687][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:30:57,187][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:30:57,689][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:30:58,191][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:30:58,693][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21743 tokens. [2026-03-26 03:30:59,331][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.53%, ΔTime: 00:01:04 [2026-03-26 03:31:00,065][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:31:00,067][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:31:00,069][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:31:00,718][__main__][INFO] - Iteration 327 took 1m 14s (9.51% Gen, 89.62% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 55h 8m 15s. Estimated total time: 62h 4m 34s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 45s. [2026-03-26 03:31:00,720][__main__][INFO] - Starting iteration 327. [2026-03-26 03:31:01,120][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:31:01,121][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:31:03,940][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:31:07,945][__main__][INFO] - Number of regex retries in iteration 327: 1 [2026-03-26 03:31:07,945][__main__][INFO] - agents played in iteration 327 are Alice, Bob [2026-03-26 03:31:08,824][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:31:08,869][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:31:08,869][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:31:08,870][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:31:09,483][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:31:09,939][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:31:10,444][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:31:10,945][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:31:11,446][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:31:11,947][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:31:12,446][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:31:12,947][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:31:13,447][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:31:13,945][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:31:14,443][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:31:14,940][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:31:15,440][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:31:15,940][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:31:16,438][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:31:16,934][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:31:17,432][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:31:17,932][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:31:18,434][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:31:18,934][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:31:19,433][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:31:19,933][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:31:20,432][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:31:20,931][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:31:21,431][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:31:21,930][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:31:22,428][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:31:22,926][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:31:23,425][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:31:23,922][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:31:24,419][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:31:24,917][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:31:25,416][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:31:25,916][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:31:26,414][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:31:26,913][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:31:27,412][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:31:27,911][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:31:28,410][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:31:28,909][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:31:29,409][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:31:29,909][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:31:30,409][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:31:30,908][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:31:31,406][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:31:31,905][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:31:32,403][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:31:32,901][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:31:33,400][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:31:33,896][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:31:34,394][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:31:34,891][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:31:35,388][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:31:35,885][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:31:36,382][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:31:36,879][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:31:37,377][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:31:37,875][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:31:38,374][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:31:38,872][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:31:39,370][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:31:39,868][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:31:40,366][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:31:40,863][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:31:41,360][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:31:41,858][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:31:42,358][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:31:42,857][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:31:43,357][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:31:43,858][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:31:44,357][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:31:44,855][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:31:45,355][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:31:45,853][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:31:46,352][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:31:46,851][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:31:47,349][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:31:47,848][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:31:48,348][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:31:48,847][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:31:49,347][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:31:49,846][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:31:50,349][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:31:50,848][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:31:51,348][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:31:51,848][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:31:52,347][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:31:52,847][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:31:53,347][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:31:53,846][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:31:54,345][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:31:54,844][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:31:55,344][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:31:55,842][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:31:56,339][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:31:56,837][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:31:57,335][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:31:57,833][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:31:58,332][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:31:58,830][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:31:59,329][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:31:59,826][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:32:00,323][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:32:00,820][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:32:01,319][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:32:01,817][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:32:02,315][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:32:02,814][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:32:03,313][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:32:03,813][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:32:04,311][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:32:04,810][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:32:05,309][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:32:05,808][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:32:06,306][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:32:06,804][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:32:07,301][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:32:07,800][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:32:08,298][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:32:08,796][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:32:09,293][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:32:09,791][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:32:10,288][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:32:10,787][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:32:11,286][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:32:11,784][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:32:12,282][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:32:12,779][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:32:13,277][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 03:32:13,915][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 03:32:14,745][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:32:14,747][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:32:14,749][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:32:15,511][__main__][INFO] - Iteration 328 took 1m 14s (9.17% Gen, 89.80% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 2m 2s. Estimated total time: 61h 59m 35s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 59s, 500 more iterations: 10h 19m 55s. [2026-03-26 03:32:15,514][__main__][INFO] - Starting iteration 328. [2026-03-26 03:32:15,911][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:32:15,912][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:32:22,721][__main__][INFO] - Number of regex retries in iteration 328: 0 [2026-03-26 03:32:22,722][__main__][INFO] - agents played in iteration 328 are Alice, Bob [2026-03-26 03:32:23,596][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:32:23,641][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:32:23,641][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:32:23,642][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:32:24,246][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:32:24,701][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:32:25,206][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:32:25,705][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:32:26,203][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:32:26,700][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:32:27,199][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:32:27,697][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:32:28,195][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:32:28,692][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:32:29,191][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:32:29,689][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:32:30,187][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:32:30,685][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:32:31,183][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:32:31,681][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:32:32,180][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:32:32,678][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:32:33,177][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:32:33,676][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:32:34,175][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:32:34,674][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:32:35,173][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:32:35,672][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:32:36,171][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:32:36,672][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:32:37,173][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:32:37,673][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:32:38,172][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:32:38,672][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:32:39,172][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:32:39,672][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:32:40,171][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:32:40,670][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:32:41,169][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:32:41,667][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:32:42,166][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:32:42,665][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:32:43,163][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:32:43,661][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:32:44,159][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:32:44,657][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:32:45,157][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:32:45,660][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:32:46,161][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:32:46,661][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:32:47,160][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:32:47,659][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:32:48,159][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:32:48,658][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:32:49,155][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:32:49,654][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:32:50,153][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:32:50,651][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:32:51,149][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:32:51,648][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:32:52,146][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:32:52,648][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:32:53,149][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:32:53,649][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:32:54,149][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:32:54,650][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:32:55,151][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:32:55,653][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:32:56,153][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:32:56,652][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:32:57,151][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:32:57,649][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:32:58,148][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:32:58,648][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:32:59,147][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:32:59,647][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:33:00,146][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:33:00,644][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:33:01,142][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:33:01,640][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:33:02,138][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:33:02,636][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:33:03,134][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:33:03,632][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:33:04,131][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:33:04,630][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:33:05,129][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:33:05,628][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:33:06,127][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:33:06,627][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:33:07,127][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:33:07,627][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:33:08,127][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:33:08,626][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:33:09,125][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:33:09,624][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:33:10,123][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:33:10,621][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:33:11,120][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:33:11,618][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:33:12,117][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:33:12,615][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:33:13,113][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:33:13,612][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:33:14,110][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:33:14,608][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:33:15,107][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:33:15,606][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:33:16,105][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:33:16,604][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:33:17,103][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:33:17,602][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:33:18,102][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:33:18,602][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:33:19,102][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:33:19,601][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:33:20,099][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:33:20,597][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:33:21,095][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:33:21,593][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:33:22,091][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:33:22,589][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:33:23,087][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:33:23,586][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:33:24,085][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:33:24,585][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:33:25,085][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:33:25,588][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:33:26,089][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:33:26,587][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:33:27,088][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:33:27,589][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:33:28,089][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 03:33:28,723][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:33:29,474][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:33:29,476][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:33:29,478][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:33:30,129][__main__][INFO] - Iteration 329 took 1m 14s (9.18% Gen, 89.94% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 52m 7s. Estimated total time: 61h 50m 55s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 29s. [2026-03-26 03:33:30,131][__main__][INFO] - Starting iteration 329. [2026-03-26 03:33:30,535][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:33:30,536][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:33:36,801][__main__][INFO] - Number of regex retries in iteration 329: 0 [2026-03-26 03:33:36,802][__main__][INFO] - agents played in iteration 329 are Alice, Bob [2026-03-26 03:33:37,682][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:33:37,726][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:33:37,727][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:33:37,728][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:33:38,335][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:33:38,792][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:33:39,296][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:33:39,795][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:33:40,294][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:33:40,794][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:33:41,294][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:33:41,793][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:33:42,293][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:33:42,790][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:33:43,288][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:33:43,786][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:33:44,283][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:33:44,780][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:33:45,277][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:33:45,775][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:33:46,274][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:33:46,774][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:33:47,274][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:33:47,773][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:33:48,275][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:33:48,775][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:33:49,274][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:33:49,774][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:33:50,273][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:33:50,772][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:33:51,272][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:33:51,770][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:33:52,269][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:33:52,768][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:33:53,268][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:33:53,767][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:33:54,267][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:33:54,765][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:33:55,265][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:33:55,763][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:33:56,261][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:33:56,760][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:33:57,258][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:33:57,757][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:33:58,255][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:33:58,754][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:33:59,253][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:33:59,753][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:34:00,252][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:34:00,752][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:34:01,252][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:34:01,752][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:34:02,252][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:34:02,751][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:34:03,251][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:34:03,750][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:34:04,249][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:34:04,749][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:34:05,248][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:34:05,747][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:34:06,247][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:34:06,746][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:34:07,246][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:34:07,746][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:34:08,247][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:34:08,747][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:34:09,247][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:34:09,747][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:34:10,247][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:34:10,748][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:34:11,248][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:34:11,748][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:34:12,248][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:34:12,748][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:34:13,247][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:34:13,746][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:34:14,246][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:34:14,744][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:34:15,243][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:34:15,744][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:34:16,246][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:34:16,745][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:34:17,242][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:34:17,740][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:34:18,240][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:34:18,740][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:34:19,239][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:34:19,738][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:34:20,238][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:34:20,737][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:34:21,236][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:34:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:34:22,234][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:34:22,733][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:34:23,230][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:34:23,729][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:34:24,228][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:34:24,729][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:34:25,228][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:34:25,726][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:34:26,226][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:34:26,725][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:34:27,225][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:34:27,723][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:34:28,223][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:34:28,721][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:34:29,219][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:34:29,717][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:34:30,215][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:34:30,713][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:34:31,214][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:34:31,712][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:34:32,212][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:34:32,711][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:34:33,210][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:34:33,708][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:34:34,207][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:34:34,708][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:34:35,208][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:34:35,707][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:34:36,207][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:34:36,705][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:34:37,204][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:34:37,702][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:34:38,201][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:34:38,700][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:34:39,199][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:34:39,699][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:34:40,198][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:34:40,697][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:34:41,196][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:34:41,695][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:34:42,195][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 03:34:42,843][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 03:34:43,584][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:34:43,586][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:34:43,587][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:34:44,236][__main__][INFO] - Iteration 330 took 1m 13s (8.50% Gen, 90.61% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 25m 4s. Estimated total time: 61h 25m 6s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 50s, 500 more iterations: 10h 14m 11s. [2026-03-26 03:34:44,239][__main__][INFO] - Starting iteration 330. [2026-03-26 03:34:44,637][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:34:44,637][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:34:45,219][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:34:51,423][__main__][INFO] - Number of regex retries in iteration 330: 1 [2026-03-26 03:34:51,424][__main__][INFO] - agents played in iteration 330 are Alice, Bob [2026-03-26 03:34:52,302][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:34:52,346][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:34:52,347][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:34:52,348][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:34:52,949][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:34:53,406][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:34:53,910][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:34:54,409][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:34:54,909][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:34:55,408][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:34:55,907][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:34:56,407][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:34:56,907][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:34:57,406][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:34:57,905][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:34:58,403][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:34:58,902][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:34:59,400][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:34:59,899][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:35:00,398][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:35:00,897][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:35:01,396][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:35:01,895][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:35:02,394][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:35:02,893][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:35:03,391][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:35:03,891][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:35:04,391][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:35:04,890][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:35:05,389][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:35:05,889][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:35:06,388][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:35:06,887][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:35:07,386][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:35:07,887][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:35:08,387][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:35:08,886][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:35:09,386][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:35:09,884][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:35:10,382][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:35:10,880][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:35:11,378][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:35:11,876][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:35:12,374][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:35:12,873][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:35:13,373][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:35:13,872][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:35:14,369][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:35:14,868][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:35:15,366][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:35:15,865][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:35:16,365][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:35:16,864][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:35:17,363][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:35:17,862][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:35:18,363][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:35:18,862][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:35:19,360][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:35:19,859][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:35:20,357][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:35:20,856][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:35:21,359][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:35:21,862][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:35:22,363][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:35:22,866][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:35:23,369][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:35:23,871][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:35:24,373][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:35:24,876][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:35:25,376][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:35:25,876][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:35:26,375][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:35:26,875][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:35:27,373][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:35:27,873][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:35:28,373][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:35:28,873][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:35:29,371][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:35:29,871][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:35:30,370][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:35:30,869][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:35:31,368][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:35:31,866][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:35:32,364][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:35:32,863][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:35:33,362][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:35:33,861][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:35:34,359][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:35:34,859][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:35:35,358][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:35:35,857][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:35:36,356][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:35:36,855][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:35:37,354][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:35:37,854][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:35:38,353][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:35:38,853][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:35:39,353][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:35:39,853][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:35:40,353][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:35:40,853][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:35:41,352][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:35:41,850][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:35:42,348][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:35:42,847][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:35:43,345][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:35:43,843][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:35:44,340][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:35:44,838][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:35:45,338][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:35:45,840][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:35:46,340][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:35:46,838][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:35:47,337][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:35:47,835][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:35:48,335][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:35:48,835][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:35:49,334][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:35:49,833][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:35:50,332][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:35:50,831][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:35:51,330][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:35:51,830][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:35:52,328][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:35:52,829][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:35:53,331][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:35:53,833][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:35:54,336][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:35:54,836][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:35:55,339][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:35:55,840][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:35:56,343][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:35:56,845][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 03:35:57,487][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.33%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 03:35:58,321][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:35:58,323][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:35:58,324][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:35:59,111][__main__][INFO] - Iteration 331 took 1m 14s (9.11% Gen, 89.83% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 2m 26s. Estimated total time: 62h 3m 43s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 7s, 500 more iterations: 10h 20m 37s. [2026-03-26 03:35:59,113][__main__][INFO] - Starting iteration 331. [2026-03-26 03:35:59,511][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:35:59,512][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:36:00,106][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:36:05,742][__main__][INFO] - Number of regex retries in iteration 331: 1 [2026-03-26 03:36:05,743][__main__][INFO] - agents played in iteration 331 are Alice, Bob [2026-03-26 03:36:06,624][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:36:06,669][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:36:06,670][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:36:06,670][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:36:07,550][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:36:08,006][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:36:08,510][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:36:09,009][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:36:09,508][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:36:10,008][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:36:10,507][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:36:11,006][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:36:11,505][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:36:12,004][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:36:12,503][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:36:13,002][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:36:13,501][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:36:14,000][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:36:14,498][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:36:14,997][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:36:15,495][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:36:15,994][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:36:16,493][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:36:16,992][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:36:17,491][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:36:17,990][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:36:18,491][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:36:18,990][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:36:19,488][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:36:19,988][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:36:20,487][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:36:20,985][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:36:21,483][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:36:21,981][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:36:22,479][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:36:22,976][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:36:23,475][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:36:23,975][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:36:24,475][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:36:24,975][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:36:25,475][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:36:25,974][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:36:26,474][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:36:26,973][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:36:27,473][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:36:27,975][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:36:28,476][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:36:28,975][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:36:29,476][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:36:29,975][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:36:30,475][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:36:30,976][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:36:31,476][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:36:31,975][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:36:32,473][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:36:32,972][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:36:33,470][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:36:33,969][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:36:34,468][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:36:34,969][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:36:35,468][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:36:35,968][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:36:36,468][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:36:36,967][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:36:37,465][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:36:37,964][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:36:38,463][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:36:38,963][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:36:39,461][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:36:39,960][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:36:40,460][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:36:40,959][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:36:41,459][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:36:41,957][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:36:42,457][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:36:42,958][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:36:43,457][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:36:43,956][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:36:44,455][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:36:44,954][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:36:45,453][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:36:45,953][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:36:46,453][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:36:46,951][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:36:47,450][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:36:47,950][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:36:48,452][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:36:48,952][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:36:49,451][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:36:49,950][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:36:50,450][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:36:50,949][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:36:51,448][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:36:51,946][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:36:52,444][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:36:52,941][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:36:53,438][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:36:53,936][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:36:54,435][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:36:54,933][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:36:55,432][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:36:55,931][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:36:56,431][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:36:56,931][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:36:57,429][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:36:57,930][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:36:58,430][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:36:58,929][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:36:59,428][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:36:59,928][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:37:00,428][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:37:00,928][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:37:01,430][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:37:01,929][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:37:02,430][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:37:02,929][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:37:03,429][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:37:03,929][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:37:04,429][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:37:04,928][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:37:05,427][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:37:05,926][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:37:06,426][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:37:06,924][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:37:07,422][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:37:07,921][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:37:08,420][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:37:08,919][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:37:09,418][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:37:09,917][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:37:10,417][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:37:10,916][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:37:11,416][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:37:12,047][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:37:12,789][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:37:12,791][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:37:12,793][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:37:13,506][__main__][INFO] - Iteration 332 took 1m 13s (8.42% Gen, 90.61% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 54h 37m 17s. Estimated total time: 61h 39m 48s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 19s, 500 more iterations: 10h 16m 38s. [2026-03-26 03:37:13,509][__main__][INFO] - Starting iteration 332. [2026-03-26 03:37:13,909][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:37:13,910][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:37:20,162][__main__][INFO] - Number of regex retries in iteration 332: 0 [2026-03-26 03:37:20,162][__main__][INFO] - agents played in iteration 332 are Alice, Bob [2026-03-26 03:37:21,035][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:37:21,079][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:37:21,080][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:37:21,081][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:37:21,687][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:37:22,142][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:37:22,647][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:37:23,146][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:37:23,645][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:37:24,143][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:37:24,642][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:37:25,142][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:37:25,641][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:37:26,139][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:37:26,638][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:37:27,138][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:37:27,636][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:37:28,135][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:37:28,633][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:37:29,132][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:37:29,631][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:37:30,131][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:37:30,630][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:37:31,129][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:37:31,627][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:37:32,126][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:37:32,625][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:37:33,124][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:37:33,623][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:37:34,125][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:37:34,626][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:37:35,125][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:37:35,625][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:37:36,124][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:37:36,624][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:37:37,125][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:37:37,626][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:37:38,124][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:37:38,622][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:37:39,120][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:37:39,618][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:37:40,116][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:37:40,614][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:37:41,113][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:37:41,611][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:37:42,110][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:37:42,609][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:37:43,108][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:37:43,608][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:37:44,106][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:37:44,605][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:37:45,105][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:37:45,604][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:37:46,103][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:37:46,602][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:37:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:37:47,600][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:37:48,100][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:37:48,600][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:37:49,099][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:37:49,598][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:37:50,097][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:37:50,597][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:37:51,096][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:37:51,595][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:37:52,094][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:37:52,594][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:37:53,094][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:37:53,594][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:37:54,093][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:37:54,592][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:37:55,091][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:37:55,591][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:37:56,090][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:37:56,590][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:37:57,089][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:37:57,588][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:37:58,087][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:37:58,586][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:37:59,084][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:37:59,582][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:38:00,080][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:38:00,579][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:38:01,078][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:38:01,576][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:38:02,076][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:38:02,576][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:38:03,075][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:38:03,575][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:38:04,074][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:38:04,574][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:38:05,074][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:38:05,573][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:38:06,073][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:38:06,574][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:38:07,075][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:38:07,575][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:38:08,076][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:38:08,576][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:38:09,076][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:38:09,577][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:38:10,075][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:38:10,574][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:38:11,073][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:38:11,572][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:38:12,071][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:38:12,570][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:38:13,068][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:38:13,567][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:38:14,066][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:38:14,565][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:38:15,064][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:38:15,563][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:38:16,062][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:38:16,562][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:38:17,061][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:38:17,561][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:38:18,061][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:38:18,562][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:38:19,062][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:38:19,561][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:38:20,061][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:38:20,561][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:38:21,062][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:38:21,560][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:38:22,059][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:38:22,559][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:38:23,058][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:38:23,559][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:38:24,058][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:38:24,557][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:38:25,056][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:38:25,556][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:38:26,191][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:38:26,936][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:38:26,938][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:38:26,940][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:38:27,633][__main__][INFO] - Iteration 333 took 1m 13s (8.48% Gen, 90.58% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 22m 29s. Estimated total time: 61h 26m 14s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 52s, 500 more iterations: 10h 14m 22s. [2026-03-26 03:38:27,635][__main__][INFO] - Starting iteration 333. [2026-03-26 03:38:28,041][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:38:28,042][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:38:28,623][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:38:34,178][__main__][INFO] - Number of regex retries in iteration 333: 1 [2026-03-26 03:38:34,179][__main__][INFO] - agents played in iteration 333 are Alice, Bob [2026-03-26 03:38:35,054][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:38:35,098][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:38:35,099][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:38:35,100][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:38:35,708][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:38:36,163][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:38:36,666][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:38:37,165][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:38:37,663][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:38:38,161][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:38:38,660][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:38:39,159][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:38:39,658][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:38:40,154][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:38:40,649][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:38:41,144][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:38:41,639][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:38:42,134][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:38:42,629][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:38:43,125][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:38:43,621][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:38:44,121][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:38:44,620][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:38:45,118][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:38:45,620][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:38:46,120][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:38:46,619][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:38:47,117][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:38:47,615][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:38:48,117][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:38:48,619][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:38:49,117][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:38:49,616][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:38:50,114][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:38:50,613][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:38:51,112][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:38:51,611][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:38:52,109][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:38:52,607][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:38:53,105][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:38:53,602][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:38:54,099][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:38:54,597][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:38:55,095][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:38:55,593][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:38:56,092][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:38:56,592][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:38:57,091][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:38:57,590][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:38:58,089][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:38:58,588][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:38:59,086][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:38:59,587][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:39:00,087][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:39:00,587][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:39:01,088][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:39:01,587][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:39:02,086][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:39:02,584][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:39:03,083][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:39:03,582][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:39:04,101][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:39:04,606][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:39:05,107][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:39:05,609][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:39:06,110][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:39:06,612][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:39:07,112][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:39:07,613][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:39:08,111][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:39:08,608][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:39:09,107][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:39:09,606][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:39:10,104][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:39:10,602][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:39:11,099][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:39:11,597][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:39:12,092][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:39:12,586][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:39:13,080][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:39:13,576][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:39:14,072][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:39:14,567][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:39:15,062][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:39:15,557][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:39:16,056][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:39:16,554][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:39:17,053][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:39:17,552][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:39:18,053][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:39:18,554][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:39:19,052][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:39:19,551][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:39:20,051][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:39:20,550][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:39:21,048][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:39:21,547][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:39:22,046][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:39:22,545][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:39:23,044][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:39:23,543][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:39:24,042][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:39:24,540][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:39:25,038][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:39:25,536][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:39:26,035][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:39:26,533][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:39:27,031][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:39:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:39:28,029][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:39:28,528][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:39:29,027][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:39:29,525][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:39:30,023][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:39:30,522][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:39:31,020][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:39:31,518][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:39:32,018][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:39:32,517][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:39:33,016][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:39:33,515][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:39:34,013][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:39:34,512][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:39:35,012][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:39:35,512][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:39:36,013][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:39:36,515][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:39:37,016][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:39:37,516][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:39:38,018][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:39:38,519][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:39:39,021][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:39:39,522][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:39:40,154][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 03:39:40,880][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:39:40,883][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:39:40,884][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:39:41,593][__main__][INFO] - Iteration 334 took 1m 13s (8.34% Gen, 90.69% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 12m 39s. Estimated total time: 61h 17m 39s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 35s, 500 more iterations: 10h 12m 56s. [2026-03-26 03:39:41,596][__main__][INFO] - Starting iteration 334. [2026-03-26 03:39:41,998][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:39:41,998][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:39:48,282][__main__][INFO] - Number of regex retries in iteration 334: 0 [2026-03-26 03:39:48,283][__main__][INFO] - agents played in iteration 334 are Alice, Bob [2026-03-26 03:39:49,164][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:39:49,208][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:39:49,209][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:39:49,209][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:39:49,816][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:39:50,272][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:39:50,777][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:39:51,276][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:39:51,775][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:39:52,273][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:39:52,774][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:39:53,273][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:39:53,772][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:39:54,271][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:39:54,773][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:39:55,274][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:39:55,774][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:39:56,273][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:39:56,773][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:39:57,273][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:39:57,773][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:39:58,273][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:39:58,773][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:39:59,273][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:39:59,773][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:40:00,273][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:40:00,773][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:40:01,273][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:40:01,773][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:40:02,272][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:40:02,770][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:40:03,268][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:40:03,765][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:40:04,264][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:40:04,762][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:40:05,260][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:40:05,759][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:40:06,258][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:40:06,758][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:40:07,258][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:40:07,757][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:40:08,257][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:40:08,756][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:40:09,257][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:40:09,757][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:40:10,256][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:40:10,755][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:40:11,254][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:40:11,753][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:40:12,253][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:40:12,752][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:40:13,251][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:40:13,750][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:40:14,249][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:40:14,747][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:40:15,246][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:40:15,751][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:40:16,251][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:40:16,750][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:40:17,248][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:40:17,747][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:40:18,247][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:40:18,745][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:40:19,243][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:40:19,741][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:40:20,239][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:40:20,738][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:40:21,236][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:40:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:40:22,234][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:40:22,734][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:40:23,233][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:40:23,733][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:40:24,232][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:40:24,731][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:40:25,230][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:40:25,728][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:40:26,227][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:40:26,726][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:40:27,226][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:40:27,727][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:40:28,228][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:40:28,728][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:40:29,228][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:40:29,728][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:40:30,228][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:40:30,731][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:40:31,231][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:40:31,731][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:40:32,230][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:40:32,732][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:40:33,232][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:40:33,733][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:40:34,231][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:40:34,730][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:40:35,227][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:40:35,725][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:40:36,222][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:40:36,720][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:40:37,218][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:40:37,715][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:40:38,214][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:40:38,714][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:40:39,212][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:40:39,712][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:40:40,211][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:40:40,711][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:40:41,210][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:40:41,709][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:40:42,210][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:40:42,710][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:40:43,209][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:40:43,709][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:40:44,207][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:40:44,705][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:40:45,203][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:40:45,701][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:40:46,200][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:40:46,698][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:40:47,197][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:40:47,696][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:40:48,197][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:40:48,696][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:40:49,194][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:40:49,693][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:40:50,193][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:40:50,691][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:40:51,190][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:40:51,688][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:40:52,186][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:40:52,684][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:40:53,181][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:40:53,679][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21731 tokens. [2026-03-26 03:40:54,310][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:40:55,049][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:40:55,051][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:40:55,053][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:40:55,746][__main__][INFO] - Iteration 335 took 1m 13s (8.52% Gen, 90.54% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 21m 12s. Estimated total time: 61h 27m 25s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 54s, 500 more iterations: 10h 14m 34s. [2026-03-26 03:40:55,748][__main__][INFO] - Starting iteration 335. [2026-03-26 03:40:56,147][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:40:56,148][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:41:02,355][__main__][INFO] - Number of regex retries in iteration 335: 0 [2026-03-26 03:41:02,356][__main__][INFO] - agents played in iteration 335 are Alice, Bob [2026-03-26 03:41:03,234][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:41:03,278][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:41:03,279][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:41:03,280][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:41:03,886][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:41:04,341][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:41:04,845][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:41:05,344][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:41:05,842][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:41:06,340][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:41:06,839][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:41:07,338][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:41:07,836][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:41:08,337][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:41:08,836][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:41:09,335][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:41:09,835][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:41:10,334][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:41:10,833][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:41:11,332][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:41:11,832][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:41:12,331][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:41:12,829][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:41:13,327][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:41:13,826][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:41:14,324][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:41:14,822][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:41:15,321][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:41:15,819][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:41:16,315][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:41:16,811][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:41:17,305][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:41:17,800][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:41:18,298][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:41:18,794][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:41:19,289][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:41:19,783][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:41:20,278][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:41:20,774][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:41:21,270][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:41:21,766][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:41:22,261][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:41:22,756][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:41:23,252][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:41:23,748][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:41:24,248][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:41:24,748][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:41:25,248][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:41:25,750][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:41:26,250][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:41:26,750][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:41:27,250][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:41:27,749][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:41:28,249][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:41:28,750][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:41:29,249][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:41:29,749][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:41:30,248][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:41:30,747][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:41:31,246][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:41:31,744][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:41:32,243][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:41:32,741][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:41:33,241][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:41:33,740][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:41:34,239][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:41:34,738][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:41:35,236][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:41:35,735][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:41:36,234][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:41:36,733][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:41:37,232][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:41:37,731][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:41:38,229][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:41:38,727][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:41:39,224][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:41:39,722][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:41:40,221][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:41:40,718][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:41:41,218][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:41:41,717][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:41:42,216][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:41:42,714][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:41:43,213][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:41:43,712][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:41:44,211][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:41:44,710][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:41:45,210][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:41:45,709][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:41:46,208][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:41:46,707][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:41:47,205][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:41:47,702][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:41:48,198][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:41:48,696][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:41:49,192][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:41:49,688][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:41:50,183][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:41:50,678][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:41:51,174][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:41:51,670][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:41:52,164][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:41:52,659][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:41:53,154][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:41:53,649][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:41:54,143][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:41:54,636][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:41:55,133][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:41:55,629][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:41:56,129][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:41:56,631][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:41:57,130][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:41:57,630][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:41:58,128][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:41:58,627][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:41:59,126][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:41:59,625][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:42:00,124][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:42:00,622][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:42:01,121][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:42:01,619][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:42:02,117][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:42:02,615][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:42:03,113][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:42:03,612][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:42:04,112][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:42:04,611][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:42:05,110][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:42:05,608][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:42:06,106][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:42:06,604][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:42:07,102][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:42:07,601][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:42:08,244][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:42:08,998][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:42:09,000][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:42:09,001][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:42:09,686][__main__][INFO] - Iteration 336 took 1m 13s (8.44% Gen, 90.63% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 9m 30s. Estimated total time: 61h 16m 58s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 33s, 500 more iterations: 10h 12m 49s. [2026-03-26 03:42:09,688][__main__][INFO] - Starting iteration 336. [2026-03-26 03:42:10,087][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:42:10,087][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:42:11,743][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:42:16,218][__main__][INFO] - Number of regex retries in iteration 336: 1 [2026-03-26 03:42:16,219][__main__][INFO] - agents played in iteration 336 are Alice, Bob [2026-03-26 03:42:17,093][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:42:17,137][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:42:17,138][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:42:17,138][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:42:17,739][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:42:18,193][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:42:18,697][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:42:19,197][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:42:19,696][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:42:20,195][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:42:20,694][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:42:21,193][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:42:21,693][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:42:22,194][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:42:22,693][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:42:23,191][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:42:23,690][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:42:24,190][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:42:24,690][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:42:25,191][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:42:25,691][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:42:26,190][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:42:26,688][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:42:27,187][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:42:27,685][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:42:28,184][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:42:28,682][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:42:29,181][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:42:29,681][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:42:30,183][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:42:30,682][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:42:31,182][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:42:31,682][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:42:32,182][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:42:32,682][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:42:33,182][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:42:33,682][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:42:34,179][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:42:34,677][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:42:35,176][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:42:35,674][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:42:36,172][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:42:36,672][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:42:37,171][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:42:37,669][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:42:38,168][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:42:38,666][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:42:39,164][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:42:39,663][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:42:40,162][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:42:40,661][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:42:41,160][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:42:41,659][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:42:42,158][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:42:42,657][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:42:43,156][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:42:43,656][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:42:44,156][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:42:44,655][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:42:45,154][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:42:45,653][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:42:46,152][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:42:46,651][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:42:47,151][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:42:47,650][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:42:48,150][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:42:48,650][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:42:49,150][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:42:49,648][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:42:50,146][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:42:50,644][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:42:51,141][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:42:51,640][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:42:52,138][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:42:52,637][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:42:53,135][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:42:53,634][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:42:54,133][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:42:54,633][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:42:55,133][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:42:55,632][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:42:56,131][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:42:56,631][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:42:57,130][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:42:57,629][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:42:58,127][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:42:58,626][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:42:59,125][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:42:59,623][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:43:00,122][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:43:00,621][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:43:01,120][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:43:01,618][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:43:02,118][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:43:02,617][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:43:03,117][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:43:03,618][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:43:04,118][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:43:04,618][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:43:05,118][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:43:05,618][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:43:06,115][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:43:06,614][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:43:07,113][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:43:07,612][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:43:08,111][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:43:08,609][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:43:09,107][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:43:09,605][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:43:10,104][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:43:10,603][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:43:11,102][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:43:11,601][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:43:12,101][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:43:12,599][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:43:13,098][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:43:13,597][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:43:14,096][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:43:14,595][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:43:15,093][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:43:15,593][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:43:16,091][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:43:16,590][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:43:17,088][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:43:17,587][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:43:18,085][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:43:18,585][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:43:19,084][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:43:19,582][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:43:20,081][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:43:20,579][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:43:21,077][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:43:21,576][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21718 tokens. [2026-03-26 03:43:22,200][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 03:43:22,941][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:43:22,943][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:43:22,945][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:43:23,674][__main__][INFO] - Iteration 337 took 1m 13s (8.33% Gen, 90.68% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 10m 40s. Estimated total time: 61h 19m 22s. Time estimates for 10 more iterations: 12m 15s, 100 more iterations: 2h 2m 38s, 500 more iterations: 10h 13m 13s. [2026-03-26 03:43:23,676][__main__][INFO] - Starting iteration 337. [2026-03-26 03:43:24,078][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:43:24,079][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:43:30,971][__main__][INFO] - Number of regex retries in iteration 337: 0 [2026-03-26 03:43:30,972][__main__][INFO] - agents played in iteration 337 are Alice, Bob [2026-03-26 03:43:31,852][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:43:31,896][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:43:31,897][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:43:31,898][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:43:32,506][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:43:32,963][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:43:33,467][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:43:33,966][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:43:34,465][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:43:34,964][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:43:35,464][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:43:35,963][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:43:36,463][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:43:36,962][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:43:37,460][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:43:37,959][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:43:38,457][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:43:38,956][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:43:39,454][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:43:39,953][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:43:40,452][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:43:40,950][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:43:41,449][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:43:41,948][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:43:42,446][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:43:42,946][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:43:43,445][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:43:43,943][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:43:44,442][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:43:44,940][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:43:45,438][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:43:45,937][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:43:46,436][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:43:46,935][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:43:47,433][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:43:47,933][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:43:48,433][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:43:48,932][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:43:49,431][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:43:49,931][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:43:50,429][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:43:50,929][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:43:51,429][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:43:51,929][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:43:52,428][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:43:52,925][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:43:53,423][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:43:53,920][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:43:54,418][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:43:54,917][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:43:55,416][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:43:55,914][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:43:56,412][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:43:56,911][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:43:57,409][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:43:57,907][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:43:58,404][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:43:58,902][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:43:59,401][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:43:59,900][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:44:00,399][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:44:00,898][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:44:01,397][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:44:01,896][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:44:02,395][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:44:02,895][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:44:03,394][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:44:03,894][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:44:04,393][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:44:04,892][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:44:05,390][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:44:05,889][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:44:06,387][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:44:06,885][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:44:07,383][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:44:07,882][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:44:08,380][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:44:08,879][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:44:09,377][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:44:09,876][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:44:10,376][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:44:10,875][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:44:11,375][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:44:11,874][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:44:12,374][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:44:12,873][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:44:13,372][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:44:13,871][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:44:14,370][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:44:14,872][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:44:15,372][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:44:15,872][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:44:16,373][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:44:16,871][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:44:17,369][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:44:17,868][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:44:18,369][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:44:18,869][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:44:19,367][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:44:19,866][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:44:20,364][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:44:20,864][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:44:21,363][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:44:21,862][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:44:22,362][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:44:22,861][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:44:23,361][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:44:23,860][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:44:24,359][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:44:24,858][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:44:25,357][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:44:25,856][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:44:26,355][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:44:26,854][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:44:27,352][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:44:27,852][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:44:28,350][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:44:28,849][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:44:29,347][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:44:29,846][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:44:30,345][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:44:30,843][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:44:31,342][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:44:31,841][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:44:32,339][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:44:32,838][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:44:33,337][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:44:33,836][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:44:34,337][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:44:34,837][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:44:35,336][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:44:35,835][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:44:36,335][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 03:44:36,978][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 03:44:37,713][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:44:37,716][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:44:37,717][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:44:38,457][__main__][INFO] - Iteration 338 took 1m 14s (9.27% Gen, 89.74% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 49m 0s. Estimated total time: 61h 58m 56s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 57s, 500 more iterations: 10h 19m 49s. [2026-03-26 03:44:38,459][__main__][INFO] - Starting iteration 338. [2026-03-26 03:44:38,855][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:44:38,856][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:44:45,192][__main__][INFO] - Number of regex retries in iteration 338: 0 [2026-03-26 03:44:45,193][__main__][INFO] - agents played in iteration 338 are Alice, Bob [2026-03-26 03:44:46,070][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:44:46,114][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:44:46,115][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:44:46,115][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:44:46,724][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:44:47,181][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:44:47,686][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:44:48,189][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:44:48,693][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:44:49,194][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:44:49,694][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:44:50,194][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:44:50,695][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:44:51,193][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:44:51,693][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:44:52,191][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:44:52,690][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:44:53,188][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:44:53,686][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:44:54,183][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:44:54,680][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:44:55,179][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:44:55,678][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:44:56,177][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:44:56,676][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:44:57,175][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:44:57,673][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:44:58,172][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:44:58,670][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:44:59,169][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:44:59,667][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:45:00,166][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:45:00,664][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:45:01,163][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:45:01,662][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:45:02,161][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:45:02,659][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:45:03,159][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:45:03,659][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:45:04,158][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:45:04,658][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:45:05,158][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:45:05,657][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:45:06,156][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:45:06,656][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:45:07,156][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:45:07,657][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:45:08,157][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:45:08,657][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:45:09,158][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:45:09,658][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:45:10,158][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:45:10,658][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:45:11,155][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:45:11,651][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:45:12,146][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:45:12,642][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:45:13,137][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:45:13,631][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:45:14,126][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:45:14,621][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:45:15,120][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:45:15,618][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:45:16,117][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:45:16,616][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:45:17,115][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:45:17,614][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:45:18,113][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:45:18,614][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:45:19,114][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:45:19,613][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:45:20,114][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:45:20,613][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:45:21,113][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:45:21,612][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:45:22,112][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:45:22,611][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:45:23,109][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:45:23,607][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:45:24,106][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:45:24,603][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:45:25,101][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:45:25,599][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:45:26,097][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:45:26,595][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:45:27,094][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:45:27,592][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:45:28,092][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:45:28,592][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:45:29,091][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:45:29,591][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:45:30,089][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:45:30,587][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:45:31,085][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:45:31,584][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:45:32,083][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:45:32,582][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:45:33,081][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:45:33,579][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:45:34,079][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:45:34,578][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:45:35,079][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:45:35,578][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:45:36,077][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:45:36,579][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:45:37,079][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:45:37,578][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:45:38,078][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:45:38,578][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:45:39,078][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:45:39,578][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:45:40,077][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:45:40,577][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:45:41,077][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:45:41,576][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:45:42,075][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:45:42,574][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:45:43,071][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:45:43,567][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:45:44,062][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:45:44,557][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:45:45,053][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:45:45,549][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:45:46,044][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:45:46,539][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:45:47,037][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:45:47,536][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:45:48,036][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:45:48,537][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:45:49,036][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:45:49,534][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:45:50,032][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:45:50,530][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 03:45:51,162][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:45:51,963][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:45:51,965][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:45:51,967][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:45:52,700][__main__][INFO] - Iteration 339 took 1m 13s (8.58% Gen, 90.42% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 21m 3s. Estimated total time: 61h 32m 14s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 4s, 500 more iterations: 10h 15m 22s. [2026-03-26 03:45:52,702][__main__][INFO] - Starting iteration 339. [2026-03-26 03:45:53,101][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:45:53,101][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:45:59,537][__main__][INFO] - Number of regex retries in iteration 339: 0 [2026-03-26 03:45:59,538][__main__][INFO] - agents played in iteration 339 are Alice, Bob [2026-03-26 03:46:00,411][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:46:00,455][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:46:00,456][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:46:00,457][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:46:01,061][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:46:01,517][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:46:02,021][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:46:02,520][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:46:03,020][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:46:03,520][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:46:04,020][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:46:04,519][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:46:05,019][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:46:05,518][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:46:06,015][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:46:06,514][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:46:07,014][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:46:07,512][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:46:08,012][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:46:08,511][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:46:09,009][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:46:09,509][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:46:10,007][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:46:10,505][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:46:11,003][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:46:11,502][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:46:12,000][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:46:12,498][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:46:12,997][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:46:13,496][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:46:13,995][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:46:14,494][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:46:14,994][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:46:15,496][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:46:15,998][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:46:16,497][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:46:16,996][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:46:17,495][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:46:17,995][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:46:18,496][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:46:18,995][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:46:19,494][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:46:19,993][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:46:20,491][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:46:20,988][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:46:21,484][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:46:21,981][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:46:22,479][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:46:22,976][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:46:23,474][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:46:23,971][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:46:24,468][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:46:24,966][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:46:25,463][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:46:25,959][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:46:26,458][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:46:26,955][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:46:27,453][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:46:27,950][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:46:28,446][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:46:28,944][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:46:29,443][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:46:29,940][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:46:30,438][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:46:30,937][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:46:31,435][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:46:31,934][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:46:32,432][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:46:32,932][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:46:33,430][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:46:33,929][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:46:34,427][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:46:34,926][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:46:35,425][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:46:35,923][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:46:36,424][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:46:36,923][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:46:37,421][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:46:37,919][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:46:38,416][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:46:38,914][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:46:39,413][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:46:39,913][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:46:40,412][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:46:40,910][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:46:41,409][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:46:41,907][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:46:42,405][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:46:42,904][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:46:43,402][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:46:43,901][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:46:44,400][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:46:44,899][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:46:45,397][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:46:45,895][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:46:46,392][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:46:46,889][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:46:47,386][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:46:47,883][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:46:48,383][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:46:48,882][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:46:49,381][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:46:49,879][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:46:50,377][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:46:50,875][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:46:51,374][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:46:51,872][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:46:52,372][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:46:52,871][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:46:53,367][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:46:53,863][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:46:54,359][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:46:54,856][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:46:55,352][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:46:55,849][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:46:56,345][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:46:56,841][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:46:57,338][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:46:57,836][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:46:58,333][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:46:58,830][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:46:59,327][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:46:59,823][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:47:00,319][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:47:00,817][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:47:01,316][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:47:01,815][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:47:02,315][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:47:02,814][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:47:03,313][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:47:03,812][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:47:04,310][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:47:04,809][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:47:05,435][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 03:47:06,169][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:47:06,171][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:47:06,173][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:47:06,924][__main__][INFO] - Iteration 340 took 1m 13s (8.72% Gen, 90.26% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 18m 47s. Estimated total time: 61h 31m 12s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 2s, 500 more iterations: 10h 15m 12s. [2026-03-26 03:47:06,926][__main__][INFO] - Starting iteration 340. [2026-03-26 03:47:07,327][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:47:07,327][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:47:14,018][__main__][INFO] - Number of regex retries in iteration 340: 0 [2026-03-26 03:47:14,019][__main__][INFO] - agents played in iteration 340 are Alice, Bob [2026-03-26 03:47:14,896][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:47:14,940][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:47:14,941][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:47:14,941][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:47:15,545][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:47:16,000][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:47:16,503][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:47:17,002][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:47:17,501][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:47:18,001][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:47:18,503][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:47:19,001][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:47:19,500][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:47:20,001][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:47:20,501][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:47:21,001][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:47:21,502][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:47:22,002][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:47:22,502][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:47:23,003][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:47:23,502][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:47:24,001][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:47:24,499][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:47:24,998][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:47:25,497][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:47:25,995][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:47:26,493][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:47:26,991][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:47:27,489][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:47:27,986][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:47:28,484][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:47:28,982][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:47:29,481][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:47:29,979][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:47:30,478][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:47:30,979][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:47:31,477][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:47:31,975][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:47:32,473][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:47:32,971][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:47:33,473][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:47:33,971][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:47:34,470][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:47:34,967][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:47:35,466][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:47:35,963][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:47:36,461][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:47:36,960][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:47:37,459][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:47:37,958][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:47:38,456][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:47:38,955][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:47:39,523][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:47:40,021][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:47:40,520][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:47:41,018][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:47:41,518][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:47:42,016][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:47:42,514][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:47:43,013][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:47:43,512][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:47:44,010][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:47:44,508][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:47:45,009][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:47:45,512][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:47:46,014][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:47:46,514][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:47:47,014][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:47:47,513][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:47:48,013][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:47:48,516][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:47:49,017][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:47:49,517][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:47:50,017][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:47:50,516][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:47:51,014][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:47:51,515][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:47:52,015][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:47:52,516][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:47:53,017][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:47:53,517][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:47:54,018][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:47:54,518][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:47:55,020][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:47:55,521][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:47:56,021][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:47:56,521][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:47:57,020][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:47:57,519][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:47:58,018][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:47:58,516][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:47:59,016][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:47:59,516][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:48:00,015][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:48:00,513][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:48:01,012][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:48:01,510][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:48:02,008][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:48:02,507][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:48:03,005][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:48:03,504][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:48:04,003][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:48:04,501][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:48:04,999][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:48:05,497][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:48:05,995][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:48:06,494][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:48:06,994][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:48:07,493][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:48:07,991][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:48:08,490][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:48:08,988][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:48:09,486][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:48:09,983][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:48:10,481][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:48:10,979][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:48:11,479][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:48:11,978][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:48:12,477][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:48:12,976][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:48:13,476][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:48:13,974][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:48:14,473][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:48:14,975][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:48:15,474][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:48:15,974][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:48:16,475][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:48:16,975][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:48:17,474][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:48:17,975][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:48:18,478][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:48:18,979][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:48:19,480][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 03:48:20,153][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:48:20,886][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:48:20,888][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:48:20,890][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:48:21,541][__main__][INFO] - Iteration 341 took 1m 14s (9.02% Gen, 90.10% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 37m 6s. Estimated total time: 61h 50m 45s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 27s. [2026-03-26 03:48:21,543][__main__][INFO] - Starting iteration 341. [2026-03-26 03:48:21,949][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:48:21,950][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:48:25,287][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:48:28,504][__main__][INFO] - Number of regex retries in iteration 341: 1 [2026-03-26 03:48:28,505][__main__][INFO] - agents played in iteration 341 are Alice, Bob [2026-03-26 03:48:29,392][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:48:29,718][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:48:29,718][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:48:29,719][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:48:30,323][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:48:30,779][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:48:31,283][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:48:31,782][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:48:32,282][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:48:32,781][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:48:33,285][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:48:33,785][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:48:34,285][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:48:34,789][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:48:35,292][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:48:35,795][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:48:36,297][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:48:36,800][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:48:37,303][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:48:37,806][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:48:38,310][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:48:38,812][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:48:39,313][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:48:39,813][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:48:40,315][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:48:40,816][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:48:41,317][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:48:41,816][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:48:42,316][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:48:42,822][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:48:43,321][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:48:43,819][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:48:44,316][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:48:44,814][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:48:45,311][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:48:45,808][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:48:46,304][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:48:46,804][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:48:47,308][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:48:47,809][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:48:48,310][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:48:48,812][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:48:49,312][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:48:49,812][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:48:50,311][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:48:50,811][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:48:51,312][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:48:51,811][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:48:52,311][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:48:52,810][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:48:53,310][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:48:53,809][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:48:54,308][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:48:54,809][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:48:55,309][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:48:55,809][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:48:56,308][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:48:56,808][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:48:57,307][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:48:57,808][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:48:58,308][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:48:58,807][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:48:59,306][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:48:59,805][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:49:00,303][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:49:00,802][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:49:01,300][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:49:01,799][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:49:02,298][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:49:02,796][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:49:03,294][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:49:03,792][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:49:04,289][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:49:04,787][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:49:05,284][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:49:05,781][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:49:06,279][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:49:06,780][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:49:07,280][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:49:07,780][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:49:08,281][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:49:08,781][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:49:09,280][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:49:09,780][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:49:10,281][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:49:10,782][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:49:11,280][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:49:11,778][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:49:12,276][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:49:12,773][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:49:13,271][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:49:13,771][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:49:14,270][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:49:14,765][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:49:15,262][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:49:15,760][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:49:16,256][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:49:16,751][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:49:17,246][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:49:17,740][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:49:18,237][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:49:18,737][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:49:19,236][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:49:19,735][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:49:20,234][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:49:20,732][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:49:21,230][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:49:21,728][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:49:22,226][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:49:22,724][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:49:23,222][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:49:23,721][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:49:24,219][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:49:24,717][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:49:25,216][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:49:25,714][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:49:26,211][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:49:26,709][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:49:27,207][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:49:27,706][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:49:28,205][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:49:28,704][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:49:29,203][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:49:29,701][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:49:30,199][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:49:30,698][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:49:31,196][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:49:31,695][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:49:32,194][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:49:32,693][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:49:33,192][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:49:33,690][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:49:34,188][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 03:49:34,869][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 03:49:35,606][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:49:35,608][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:49:35,610][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:49:36,324][__main__][INFO] - Iteration 342 took 1m 14s (8.81% Gen, 90.23% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 54h 43m 50s. Estimated total time: 61h 58m 44s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 57s, 500 more iterations: 10h 19m 47s. [2026-03-26 03:49:36,326][__main__][INFO] - Starting iteration 342. [2026-03-26 03:49:36,729][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:49:36,730][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:49:37,313][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:49:43,393][__main__][INFO] - Number of regex retries in iteration 342: 1 [2026-03-26 03:49:43,393][__main__][INFO] - agents played in iteration 342 are Alice, Bob [2026-03-26 03:49:44,352][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:49:44,397][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:49:44,398][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:49:44,398][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:49:45,006][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:49:45,457][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:49:45,961][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:49:46,458][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:49:46,957][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:49:47,455][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:49:47,954][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:49:48,454][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:49:48,950][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:49:49,452][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:49:49,954][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:49:50,455][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:49:50,956][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:49:51,456][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:49:51,957][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:49:52,457][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:49:52,958][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:49:53,458][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:49:53,960][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:49:54,461][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:49:54,961][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:49:55,462][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:49:55,962][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:49:56,462][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:49:56,963][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:49:57,462][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:49:57,962][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:49:58,461][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:49:58,960][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:49:59,459][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:49:59,958][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:50:00,456][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:50:00,956][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:50:01,457][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:50:01,958][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:50:02,457][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:50:02,958][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:50:03,458][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:50:03,958][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:50:04,458][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:50:04,958][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:50:05,459][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:50:05,960][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:50:06,461][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:50:06,961][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:50:07,462][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:50:07,963][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:50:08,463][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:50:08,963][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:50:09,467][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:50:09,968][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:50:10,469][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:50:10,970][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:50:11,470][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:50:11,971][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:50:12,472][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:50:12,972][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:50:13,471][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:50:13,973][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:50:14,473][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:50:14,973][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:50:15,474][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:50:15,973][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:50:16,472][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:50:16,971][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:50:17,467][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:50:17,965][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:50:18,464][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:50:18,962][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:50:19,460][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:50:19,958][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:50:20,455][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:50:20,951][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:50:21,452][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:50:21,954][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:50:22,455][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:50:22,954][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:50:23,454][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:50:23,955][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:50:24,456][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:50:24,957][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:50:25,456][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:50:25,955][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:50:26,454][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:50:26,954][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:50:27,452][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:50:27,951][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:50:28,449][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:50:28,947][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:50:29,445][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:50:29,943][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:50:30,442][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:50:30,941][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:50:31,440][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:50:31,939][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:50:32,439][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:50:32,938][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:50:33,437][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:50:33,938][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:50:34,438][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:50:34,938][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:50:35,438][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:50:35,939][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:50:36,439][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:50:36,941][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:50:37,444][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:50:37,946][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:50:38,447][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:50:38,948][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:50:39,449][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:50:39,951][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:50:40,455][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:50:40,957][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:50:41,461][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:50:41,964][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:50:42,467][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:50:42,970][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:50:43,472][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:50:43,975][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:50:44,478][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:50:44,980][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:50:45,485][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:50:45,989][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:50:46,490][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:50:46,991][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:50:47,495][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:50:47,999][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:50:48,504][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:50:49,005][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 03:50:49,686][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:50:50,484][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:50:50,487][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:50:50,488][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:50:51,297][__main__][INFO] - Iteration 343 took 1m 14s (8.94% Gen, 89.98% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 54h 52m 18s. Estimated total time: 62h 8m 27s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 16s, 500 more iterations: 10h 21m 24s. [2026-03-26 03:50:51,300][__main__][INFO] - Starting iteration 343. [2026-03-26 03:50:51,701][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:50:51,701][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:50:53,396][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:50:58,602][__main__][INFO] - Number of regex retries in iteration 343: 1 [2026-03-26 03:50:58,603][__main__][INFO] - agents played in iteration 343 are Alice, Bob [2026-03-26 03:50:59,477][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:50:59,521][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:50:59,522][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:50:59,522][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:51:00,129][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:51:00,584][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:51:01,087][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:51:01,586][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:51:02,086][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:51:02,585][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:51:03,084][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:51:03,583][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:51:04,082][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:51:04,578][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:51:05,074][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:51:05,570][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:51:06,065][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:51:06,561][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:51:07,057][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:51:07,553][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:51:08,048][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:51:08,549][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:51:09,048][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:51:09,546][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:51:10,045][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:51:10,544][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:51:11,042][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:51:11,541][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:51:12,040][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:51:12,539][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:51:13,039][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:51:13,539][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:51:14,039][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:51:14,538][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:51:15,037][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:51:15,536][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:51:16,036][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:51:16,537][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:51:17,036][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:51:17,536][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:51:18,036][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:51:18,538][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:51:19,036][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:51:19,536][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:51:20,035][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:51:20,533][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:51:21,031][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:51:21,530][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:51:22,027][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:51:22,525][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:51:23,022][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:51:23,519][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:51:24,017][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:51:24,520][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:51:25,021][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:51:25,521][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:51:26,022][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:51:26,522][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:51:27,022][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:51:27,522][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:51:28,021][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:51:28,521][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:51:29,020][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:51:29,519][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:51:30,018][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:51:30,516][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:51:31,017][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:51:31,517][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:51:32,016][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:51:32,516][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:51:33,015][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:51:33,514][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:51:34,011][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:51:34,510][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:51:35,009][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:51:35,506][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:51:36,004][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:51:36,500][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:51:36,995][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:51:37,490][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:51:37,984][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:51:38,480][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:51:38,975][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:51:39,470][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:51:39,965][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:51:40,464][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:51:40,964][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:51:41,462][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:51:41,961][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:51:42,461][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:51:42,961][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:51:43,460][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:51:43,959][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:51:44,458][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:51:44,957][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:51:45,455][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:51:45,956][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:51:46,456][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:51:46,955][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:51:47,454][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:51:47,954][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:51:48,455][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:51:48,955][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:51:49,455][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:51:49,955][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:51:50,455][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:51:50,955][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:51:51,454][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:51:51,956][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:51:52,454][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:51:52,955][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:51:53,454][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:51:53,953][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:51:54,452][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:51:54,950][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:51:55,448][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:51:55,946][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:51:56,446][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:51:56,946][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:51:57,446][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:51:57,947][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:51:58,447][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:51:58,947][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:51:59,448][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:51:59,948][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:52:00,447][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:52:00,946][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:52:01,445][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:52:01,945][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:52:02,444][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:52:02,942][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:52:03,442][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:52:03,941][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:52:04,617][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 03:52:05,364][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:52:05,366][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:52:05,368][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:52:06,079][__main__][INFO] - Iteration 344 took 1m 14s (9.28% Gen, 89.76% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 41m 32s. Estimated total time: 61h 58m 55s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 57s, 500 more iterations: 10h 19m 49s. [2026-03-26 03:52:06,081][__main__][INFO] - Starting iteration 344. [2026-03-26 03:52:06,487][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:52:06,487][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:52:13,211][__main__][INFO] - Number of regex retries in iteration 344: 0 [2026-03-26 03:52:13,212][__main__][INFO] - agents played in iteration 344 are Alice, Bob [2026-03-26 03:52:14,089][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:52:14,134][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:52:14,134][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:52:14,135][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:52:14,743][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:52:15,199][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:52:15,706][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:52:16,207][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:52:16,707][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:52:17,206][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:52:17,705][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:52:18,205][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:52:18,703][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:52:19,204][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:52:19,704][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:52:20,204][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:52:20,703][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:52:21,202][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:52:21,701][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:52:22,200][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:52:22,700][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:52:23,198][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:52:23,696][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:52:24,195][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:52:24,693][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:52:25,191][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:52:25,688][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:52:26,186][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:52:26,684][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:52:27,182][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:52:27,680][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:52:28,179][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:52:28,677][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:52:29,175][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:52:29,672][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:52:30,170][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:52:30,667][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:52:31,166][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:52:31,665][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:52:32,164][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:52:32,663][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:52:33,162][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:52:33,660][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:52:34,160][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:52:34,659][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:52:35,156][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:52:35,656][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:52:36,155][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:52:36,654][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:52:37,152][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:52:37,649][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:52:38,147][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:52:38,644][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:52:39,142][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:52:39,641][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:52:40,141][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:52:40,641][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:52:41,139][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:52:41,637][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:52:42,138][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:52:42,637][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:52:43,136][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:52:43,635][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:52:44,134][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:52:44,633][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:52:45,131][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:52:45,629][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:52:46,127][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:52:46,625][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:52:47,122][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:52:47,621][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:52:48,122][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:52:48,623][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:52:49,125][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:52:49,624][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:52:50,123][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:52:50,621][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:52:51,120][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:52:51,621][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:52:52,120][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:52:52,620][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:52:53,120][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:52:53,619][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:52:54,119][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:52:54,619][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:52:55,118][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:52:55,618][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:52:56,117][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:52:56,616][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:52:57,114][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:52:57,612][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:52:58,111][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:52:58,610][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:52:59,108][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:52:59,606][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:53:00,104][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:53:00,602][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:53:01,100][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:53:01,597][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:53:02,096][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:53:02,595][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:53:03,094][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:53:03,594][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:53:04,093][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:53:04,593][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:53:05,091][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:53:05,589][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:53:06,087][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:53:06,586][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:53:07,083][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:53:07,581][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:53:08,079][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:53:08,577][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:53:09,075][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:53:09,572][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:53:10,070][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:53:10,567][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:53:11,065][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:53:11,564][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:53:12,064][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:53:12,564][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:53:13,063][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:53:13,563][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:53:14,062][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:53:14,561][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:53:15,060][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:53:15,558][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:53:16,058][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:53:16,556][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:53:17,055][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:53:17,553][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:53:18,053][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:53:18,553][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:53:19,232][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 03:53:19,970][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:53:19,972][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:53:19,974][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:53:20,625][__main__][INFO] - Iteration 345 took 1m 14s (9.07% Gen, 90.05% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 28m 18s. Estimated total time: 61h 46m 57s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 33s, 500 more iterations: 10h 17m 49s. [2026-03-26 03:53:20,628][__main__][INFO] - Starting iteration 345. [2026-03-26 03:53:21,026][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:53:21,027][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:53:27,658][__main__][INFO] - Number of regex retries in iteration 345: 0 [2026-03-26 03:53:27,659][__main__][INFO] - agents played in iteration 345 are Alice, Bob [2026-03-26 03:53:28,550][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:53:28,595][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:53:28,595][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:53:28,596][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:53:29,200][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:53:29,658][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:53:30,162][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:53:30,663][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:53:31,164][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:53:31,664][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:53:32,165][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:53:32,666][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:53:33,166][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:53:33,665][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:53:34,164][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:53:34,664][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:53:35,164][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:53:35,664][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:53:36,163][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:53:36,662][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:53:37,162][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:53:37,661][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:53:38,160][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:53:38,659][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:53:39,158][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:53:39,657][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:53:40,156][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:53:40,656][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:53:41,156][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:53:41,656][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:53:42,156][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:53:42,656][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:53:43,155][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:53:43,654][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:53:44,152][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:53:44,651][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:53:45,149][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:53:45,652][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:53:46,154][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:53:46,654][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:53:47,153][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:53:47,652][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:53:48,151][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:53:48,651][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:53:49,153][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:53:49,652][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:53:50,150][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:53:50,650][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:53:51,148][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:53:51,648][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:53:52,146][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:53:52,645][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:53:53,143][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:53:53,642][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:53:54,141][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:53:54,640][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:53:55,140][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:53:55,639][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:53:56,138][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:53:56,637][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:53:57,136][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:53:57,637][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:53:58,136][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:53:58,637][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:53:59,136][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:53:59,637][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:54:00,137][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:54:00,637][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:54:01,137][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:54:01,638][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:54:02,137][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:54:02,637][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:54:03,138][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:54:03,638][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:54:04,138][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:54:04,639][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:54:05,138][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:54:05,637][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:54:06,137][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:54:06,637][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:54:07,136][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:54:07,637][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:54:08,136][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:54:08,636][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:54:09,134][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:54:09,632][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:54:10,131][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:54:10,629][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:54:11,127][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:54:11,625][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:54:12,124][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:54:12,622][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:54:13,121][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:54:13,619][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:54:14,118][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:54:14,616][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:54:15,115][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:54:15,615][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:54:16,114][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:54:16,613][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:54:17,111][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:54:17,610][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:54:18,109][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:54:18,611][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:54:19,110][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:54:19,610][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:54:20,108][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:54:20,608][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:54:21,107][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:54:21,604][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:54:22,102][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:54:22,600][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:54:23,099][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:54:23,597][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:54:24,095][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:54:24,593][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:54:25,092][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:54:25,591][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:54:26,089][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:54:26,588][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:54:27,086][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:54:27,584][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:54:28,083][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:54:28,582][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:54:29,081][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:54:29,581][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:54:30,081][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:54:30,582][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:54:31,082][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:54:31,582][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:54:32,081][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:54:32,587][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:54:33,088][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 03:54:33,755][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 03:54:34,500][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:54:34,502][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:54:34,504][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:54:35,266][__main__][INFO] - Iteration 346 took 1m 14s (8.93% Gen, 90.04% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 32m 9s. Estimated total time: 61h 52m 2s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 44s, 500 more iterations: 10h 18m 40s. [2026-03-26 03:54:35,268][__main__][INFO] - Starting iteration 346. [2026-03-26 03:54:35,667][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:54:35,667][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:54:42,250][__main__][INFO] - Number of regex retries in iteration 346: 0 [2026-03-26 03:54:42,251][__main__][INFO] - agents played in iteration 346 are Alice, Bob [2026-03-26 03:54:43,126][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:54:43,171][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:54:43,172][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:54:43,172][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:54:43,781][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:54:44,236][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:54:44,740][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:54:45,239][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:54:45,738][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:54:46,238][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:54:46,737][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:54:47,237][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:54:47,736][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:54:48,240][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:54:48,741][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:54:49,241][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:54:49,741][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:54:50,240][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:54:50,740][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:54:51,240][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:54:51,740][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:54:52,241][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:54:52,745][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:54:53,245][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:54:53,744][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:54:54,244][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:54:54,744][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:54:55,246][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:54:55,746][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:54:56,244][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:54:56,743][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:54:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:54:57,741][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:54:58,241][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:54:58,740][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:54:59,239][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:54:59,738][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:55:00,236][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:55:00,735][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:55:01,233][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:55:01,731][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:55:02,231][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:55:02,730][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:55:03,228][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:55:03,727][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:55:04,225][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:55:04,724][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:55:05,223][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:55:05,721][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:55:06,221][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:55:06,721][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:55:07,219][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:55:07,718][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:55:08,215][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:55:08,713][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:55:09,211][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:55:09,708][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:55:10,206][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:55:10,703][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:55:11,201][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:55:11,700][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:55:12,198][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:55:12,696][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:55:13,193][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:55:13,691][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:55:14,189][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:55:14,687][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:55:15,185][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:55:15,686][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:55:16,186][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:55:16,687][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:55:17,185][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:55:17,683][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:55:18,184][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:55:18,683][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:55:19,182][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:55:19,682][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:55:20,181][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:55:20,679][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:55:21,177][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:55:21,677][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:55:22,175][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:55:22,676][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:55:23,175][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:55:23,676][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:55:24,177][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:55:24,677][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:55:25,178][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:55:25,678][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:55:26,178][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:55:26,679][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:55:27,179][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:55:27,679][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:55:28,179][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:55:28,678][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:55:29,177][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:55:29,680][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:55:30,180][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:55:30,681][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:55:31,182][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:55:31,682][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:55:32,182][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:55:32,683][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:55:33,183][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:55:33,683][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:55:34,183][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:55:34,686][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:55:35,187][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:55:35,687][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:55:36,188][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:55:36,688][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:55:37,189][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:55:37,689][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:55:38,190][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:55:38,690][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:55:39,195][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:55:39,698][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:55:40,199][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:55:40,700][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:55:41,201][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:55:41,702][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:55:42,204][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:55:42,704][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:55:43,205][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:55:43,705][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:55:44,208][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:55:44,709][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:55:45,210][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:55:45,711][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:55:46,211][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:55:46,711][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:55:47,212][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:55:47,712][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 03:55:48,388][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 03:55:49,184][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:55:49,187][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:55:49,188][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:55:49,989][__main__][INFO] - Iteration 347 took 1m 14s (8.86% Gen, 90.06% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 35m 0s. Estimated total time: 61h 56m 8s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 52s, 500 more iterations: 10h 19m 21s. [2026-03-26 03:55:49,991][__main__][INFO] - Starting iteration 347. [2026-03-26 03:55:50,393][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:55:50,394][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:55:57,471][__main__][INFO] - Number of regex retries in iteration 347: 0 [2026-03-26 03:55:57,472][__main__][INFO] - agents played in iteration 347 are Alice, Bob [2026-03-26 03:55:58,353][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:55:58,397][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:55:58,398][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:55:58,399][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:55:59,002][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:55:59,457][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:55:59,960][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:56:00,460][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:56:00,958][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:56:01,457][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:56:01,956][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:56:02,455][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:56:02,953][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:56:03,454][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:56:03,953][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:56:04,453][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:56:04,953][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:56:05,452][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:56:05,951][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:56:06,450][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:56:06,949][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:56:07,448][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:56:07,947][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:56:08,445][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:56:08,944][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:56:09,446][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:56:09,945][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:56:10,444][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:56:10,942][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:56:11,445][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:56:11,945][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:56:12,445][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:56:12,944][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:56:13,445][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:56:13,944][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:56:14,445][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:56:14,945][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:56:15,443][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:56:15,942][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:56:16,440][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:56:16,938][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:56:17,436][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:56:17,936][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:56:18,436][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:56:18,936][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:56:19,436][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:56:19,937][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:56:20,436][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:56:20,936][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:56:21,434][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:56:21,933][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:56:22,432][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:56:22,931][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:56:23,429][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:56:23,927][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:56:24,424][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:56:24,923][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:56:25,421][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:56:25,919][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:56:26,418][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:56:26,915][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:56:27,413][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:56:27,910][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:56:28,409][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:56:28,907][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:56:29,406][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:56:29,906][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:56:30,404][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:56:30,902][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:56:31,400][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:56:31,899][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:56:32,397][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:56:32,895][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:56:33,393][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:56:33,890][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:56:34,388][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:56:34,886][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:56:35,385][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:56:35,884][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:56:36,383][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:56:36,883][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:56:37,382][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:56:37,882][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:56:38,382][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:56:38,881][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:56:39,380][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:56:39,879][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:56:40,378][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:56:40,877][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:56:41,376][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:56:41,874][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:56:42,373][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:56:42,871][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:56:43,370][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:56:43,870][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:56:44,370][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:56:44,870][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:56:45,372][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:56:45,875][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:56:46,375][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:56:46,875][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:56:47,375][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:56:47,874][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:56:48,373][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:56:48,872][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:56:49,372][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:56:49,870][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:56:50,368][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:56:50,866][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:56:51,366][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:56:51,866][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:56:52,365][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:56:52,865][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:56:53,364][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:56:53,863][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:56:54,363][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:56:54,863][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:56:55,362][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:56:55,859][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:56:56,357][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:56:56,856][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:56:57,353][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:56:57,851][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:56:58,348][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:56:58,845][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:56:59,343][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:56:59,842][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:57:00,344][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:57:00,843][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:57:01,342][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:57:01,841][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:57:02,340][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:57:02,839][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 03:57:03,520][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:57:04,269][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:57:04,271][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:57:04,273][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:57:05,031][__main__][INFO] - Iteration 348 took 1m 14s (9.48% Gen, 89.50% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 54h 49m 34s. Estimated total time: 62h 11m 57s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 23s, 500 more iterations: 10h 21m 59s. [2026-03-26 03:57:05,034][__main__][INFO] - Starting iteration 348. [2026-03-26 03:57:05,438][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:57:05,438][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:57:06,027][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:57:12,631][__main__][INFO] - Number of regex retries in iteration 348: 1 [2026-03-26 03:57:12,631][__main__][INFO] - agents played in iteration 348 are Alice, Bob [2026-03-26 03:57:13,534][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:57:13,579][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:57:13,580][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:57:13,580][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:57:14,179][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:57:14,633][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:57:15,137][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:57:15,636][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:57:16,134][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:57:16,633][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:57:17,131][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:57:17,629][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:57:18,129][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:57:18,630][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:57:19,128][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:57:19,627][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:57:20,125][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:57:20,624][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:57:21,123][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:57:21,622][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:57:22,121][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:57:22,621][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:57:23,121][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:57:23,620][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:57:24,119][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:57:24,619][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:57:25,119][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:57:25,619][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:57:26,119][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:57:26,620][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:57:27,120][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:57:27,619][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:57:28,119][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:57:28,618][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:57:29,118][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:57:29,618][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:57:30,118][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:57:30,638][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:57:31,142][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:57:31,642][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:57:32,142][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:57:32,642][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:57:33,143][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:57:33,643][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:57:34,143][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:57:34,644][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:57:35,143][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:57:35,643][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:57:36,143][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:57:36,642][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:57:37,141][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:57:37,641][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:57:38,140][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:57:38,639][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:57:39,138][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:57:39,637][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:57:40,138][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:57:40,637][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:57:41,137][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:57:41,636][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:57:42,136][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:57:42,635][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:57:43,133][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:57:43,631][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:57:44,129][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:57:44,626][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:57:45,125][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:57:45,624][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:57:46,123][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:57:46,622][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:57:47,120][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:57:47,619][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:57:48,118][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:57:48,618][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:57:49,117][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:57:49,616][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:57:50,114][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:57:50,613][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:57:51,112][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:57:51,611][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:57:52,109][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:57:52,607][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:57:53,105][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:57:53,604][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:57:54,104][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:57:54,603][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:57:55,103][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:57:55,603][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:57:56,103][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:57:56,603][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:57:57,102][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:57:57,602][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:57:58,101][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:57:58,600][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:57:59,100][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:57:59,600][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:58:00,101][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:58:00,600][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:58:01,100][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:58:01,600][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:58:02,100][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:58:02,600][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:58:03,105][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:58:03,606][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:58:04,107][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:58:04,608][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:58:05,108][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:58:05,610][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:58:06,110][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:58:06,610][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:58:07,110][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:58:07,615][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:58:08,117][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:58:08,620][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:58:09,121][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:58:09,622][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:58:10,124][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:58:10,625][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:58:11,126][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:58:11,626][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:58:12,132][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:58:12,634][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:58:13,137][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:58:13,639][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:58:14,141][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:58:14,642][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:58:15,144][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:58:15,648][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:58:16,151][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:58:16,651][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:58:17,155][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:58:17,657][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:58:18,160][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 03:58:18,827][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:58:19,557][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:58:19,560][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:58:19,561][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:58:20,210][__main__][INFO] - Iteration 349 took 1m 14s (9.62% Gen, 89.51% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 54h 55m 0s. Estimated total time: 62h 18m 38s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 37s, 500 more iterations: 10h 23m 6s. [2026-03-26 03:58:20,212][__main__][INFO] - Starting iteration 349. [2026-03-26 03:58:20,614][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:58:20,615][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:58:25,750][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:58:27,337][__main__][INFO] - Number of regex retries in iteration 349: 1 [2026-03-26 03:58:27,338][__main__][INFO] - agents played in iteration 349 are Alice, Bob [2026-03-26 03:58:28,248][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:58:28,297][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:58:28,298][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:58:28,299][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:58:28,904][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:58:29,361][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:58:29,865][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:58:30,366][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:58:30,866][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:58:31,366][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:58:31,866][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:58:32,366][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:58:32,874][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:58:33,376][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:58:33,878][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:58:34,378][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:58:34,878][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:58:35,378][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:58:35,879][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:58:36,379][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:58:36,880][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:58:37,381][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:58:37,884][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:58:38,384][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:58:38,884][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:58:39,385][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:58:39,885][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:58:40,386][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:58:40,886][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:58:41,385][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:58:41,883][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:58:42,382][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:58:42,880][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:58:43,378][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:58:43,877][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:58:44,376][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:58:44,874][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:58:45,372][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 03:58:45,871][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 03:58:46,369][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 03:58:46,867][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 03:58:47,366][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 03:58:47,866][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 03:58:48,367][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 03:58:48,866][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 03:58:49,364][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 03:58:49,864][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 03:58:50,363][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 03:58:50,863][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 03:58:51,363][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 03:58:51,863][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 03:58:52,362][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 03:58:52,862][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 03:58:53,361][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 03:58:53,859][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 03:58:54,358][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 03:58:54,858][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 03:58:55,357][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 03:58:55,856][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 03:58:56,354][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 03:58:56,851][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 03:58:57,348][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 03:58:57,846][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 03:58:58,344][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 03:58:58,843][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 03:58:59,340][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 03:58:59,839][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 03:59:00,336][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 03:59:00,833][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 03:59:01,331][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 03:59:01,829][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 03:59:02,328][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 03:59:02,826][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 03:59:03,325][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 03:59:03,824][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 03:59:04,323][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 03:59:04,821][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 03:59:05,320][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 03:59:05,819][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 03:59:06,319][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 03:59:06,819][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 03:59:07,318][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 03:59:07,818][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 03:59:08,317][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 03:59:08,816][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 03:59:09,314][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 03:59:09,812][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 03:59:10,312][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 03:59:10,812][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 03:59:11,311][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 03:59:11,809][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 03:59:12,308][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 03:59:12,808][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 03:59:13,306][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 03:59:13,803][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 03:59:14,302][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 03:59:14,800][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 03:59:15,298][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 03:59:15,797][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 03:59:16,295][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 03:59:16,793][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 03:59:17,291][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 03:59:17,788][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 03:59:18,289][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 03:59:18,788][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 03:59:19,286][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 03:59:19,785][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 03:59:20,283][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 03:59:20,782][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 03:59:21,280][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 03:59:21,778][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 03:59:22,276][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 03:59:22,776][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 03:59:23,275][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 03:59:23,774][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 03:59:24,273][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 03:59:24,771][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 03:59:25,269][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 03:59:25,767][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 03:59:26,265][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 03:59:26,764][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 03:59:27,263][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 03:59:27,761][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 03:59:28,259][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 03:59:28,757][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 03:59:29,255][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 03:59:29,752][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 03:59:30,249][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 03:59:30,746][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 03:59:31,245][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 03:59:31,743][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 03:59:32,242][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 03:59:32,740][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 03:59:33,403][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 03:59:34,150][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 03:59:34,152][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 03:59:34,154][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 03:59:34,888][__main__][INFO] - Iteration 350 took 1m 14s (9.05% Gen, 89.96% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 28m 50s. Estimated total time: 61h 53m 43s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 57s. [2026-03-26 03:59:34,891][__main__][INFO] - Starting iteration 350. [2026-03-26 03:59:35,290][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 6 and human policies 1. [2026-03-26 03:59:35,291][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 03:59:39,682][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 03:59:41,993][__main__][INFO] - Number of regex retries in iteration 350: 1 [2026-03-26 03:59:41,994][__main__][INFO] - agents played in iteration 350 are Alice, Bob [2026-03-26 03:59:42,865][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:59:42,909][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 03:59:42,910][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 03:59:42,910][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 03:59:43,518][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 03:59:43,974][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 03:59:44,480][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 03:59:44,980][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 03:59:45,484][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 03:59:45,987][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 03:59:46,488][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 03:59:46,989][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 03:59:47,489][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 03:59:47,993][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 03:59:48,497][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 03:59:48,998][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 03:59:49,499][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 03:59:49,999][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 03:59:50,501][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 03:59:51,003][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 03:59:51,504][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 03:59:52,003][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 03:59:52,501][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 03:59:53,001][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 03:59:53,499][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 03:59:53,999][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 03:59:54,498][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 03:59:54,998][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 03:59:55,497][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 03:59:55,997][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 03:59:56,496][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 03:59:56,995][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 03:59:57,495][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 03:59:57,995][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 03:59:58,495][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 03:59:58,995][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 03:59:59,495][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 03:59:59,995][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:00:00,495][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:00:00,997][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:00:01,498][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:00:01,999][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:00:02,498][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:00:02,998][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:00:03,499][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:00:03,996][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:00:04,497][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:00:04,998][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:00:05,497][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:00:05,997][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:00:06,496][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:00:06,995][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:00:07,494][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:00:07,991][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:00:08,488][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:00:08,986][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:00:09,484][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:00:09,982][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:00:10,479][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:00:10,977][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:00:11,475][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:00:11,974][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:00:12,472][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:00:12,972][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:00:13,472][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:00:13,971][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:00:14,469][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:00:14,969][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:00:15,468][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:00:15,967][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:00:16,466][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:00:16,967][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:00:17,467][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:00:17,967][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:00:18,469][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:00:18,968][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:00:19,467][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:00:19,968][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:00:20,467][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:00:20,968][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:00:21,468][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:00:21,968][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:00:22,467][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:00:22,966][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:00:23,466][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:00:23,966][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:00:24,465][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:00:24,963][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:00:25,462][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:00:25,963][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:00:26,463][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:00:26,963][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:00:27,462][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:00:27,962][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:00:28,462][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:00:28,962][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:00:29,462][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:00:29,963][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:00:30,470][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:00:30,971][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:00:31,473][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:00:31,976][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:00:32,477][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:00:32,978][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:00:33,481][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:00:33,983][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:00:34,485][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:00:34,986][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:00:35,489][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:00:35,991][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:00:36,492][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:00:36,993][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:00:37,494][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:00:37,995][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:00:38,495][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:00:38,995][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:00:39,496][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:00:40,000][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:00:40,500][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:00:41,000][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:00:41,499][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:00:41,999][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:00:42,498][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:00:42,999][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:00:43,499][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:00:44,001][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:00:44,502][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:00:45,005][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:00:45,507][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:00:46,008][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:00:46,509][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:00:47,010][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:00:47,510][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 04:00:48,146][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:04 [2026-03-26 04:00:48,885][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:00:48,887][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:00:48,889][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:00:50,125][__main__][INFO] - Iteration 351 took 1m 14s (8.96% Gen, 89.39% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 55m 37s. Estimated total time: 62h 21m 45s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 43s, 500 more iterations: 10h 23m 37s. [2026-03-26 04:00:50,127][__main__][INFO] - Starting iteration 351. [2026-03-26 04:00:50,529][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:00:50,530][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:00:57,167][__main__][INFO] - Number of regex retries in iteration 351: 0 [2026-03-26 04:00:57,167][__main__][INFO] - agents played in iteration 351 are Alice, Bob [2026-03-26 04:00:58,054][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:00:58,099][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:00:58,100][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:00:58,100][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:00:58,710][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:00:59,168][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:00:59,671][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:01:00,172][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:01:00,671][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:01:01,170][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:01:01,668][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:01:02,166][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:01:02,665][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:01:03,164][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:01:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:01:04,161][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:01:04,659][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:01:05,157][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:01:05,654][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:01:06,152][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:01:06,649][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:01:07,148][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:01:07,646][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:01:08,145][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:01:08,644][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:01:09,142][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:01:09,641][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:01:10,139][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:01:10,638][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:01:11,139][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:01:11,638][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:01:12,137][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:01:12,637][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:01:13,135][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:01:13,633][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:01:14,132][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:01:14,632][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:01:15,130][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:01:15,628][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:01:16,127][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:01:16,629][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:01:17,129][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:01:17,627][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:01:18,126][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:01:18,627][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:01:19,126][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:01:19,625][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:01:20,124][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:01:20,622][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:01:21,120][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:01:21,618][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:01:22,115][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:01:22,614][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:01:23,113][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:01:23,611][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:01:24,109][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:01:24,607][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:01:25,106][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:01:25,603][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:01:26,102][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:01:26,600][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:01:27,099][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:01:27,598][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:01:28,098][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:01:28,598][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:01:29,096][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:01:29,594][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:01:30,092][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:01:30,592][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:01:31,091][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:01:31,589][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:01:32,087][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:01:32,586][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:01:33,085][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:01:33,583][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:01:34,082][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:01:34,581][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:01:35,080][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:01:35,578][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:01:36,077][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:01:36,576][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:01:37,074][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:01:37,572][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:01:38,069][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:01:38,567][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:01:39,066][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:01:39,565][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:01:40,063][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:01:40,562][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:01:41,060][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:01:41,559][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:01:42,057][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:01:42,557][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:01:43,057][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:01:43,557][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:01:44,056][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:01:44,555][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:01:45,053][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:01:45,552][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:01:46,050][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:01:46,548][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:01:47,047][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:01:47,545][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:01:48,044][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:01:48,545][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:01:49,044][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:01:49,542][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:01:50,040][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:01:50,539][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:01:51,038][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:01:51,537][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:01:52,036][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:01:52,535][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:01:53,033][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:01:53,532][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:01:54,030][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:01:54,528][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:01:55,027][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:01:55,526][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:01:56,026][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:01:56,525][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:01:57,024][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:01:57,524][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:01:58,023][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:01:58,522][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:01:59,021][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:01:59,520][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:02:00,018][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:02:00,516][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:02:01,015][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:02:01,515][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:02:02,014][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:02:02,512][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 04:02:03,180][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 04:02:03,923][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:02:03,925][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:02:03,927][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:02:04,692][__main__][INFO] - Iteration 352 took 1m 14s (8.95% Gen, 90.02% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 20m 46s. Estimated total time: 61h 48m 8s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 36s, 500 more iterations: 10h 18m 1s. [2026-03-26 04:02:04,694][__main__][INFO] - Starting iteration 352. [2026-03-26 04:02:05,092][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:02:05,093][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:02:11,859][__main__][INFO] - Number of regex retries in iteration 352: 0 [2026-03-26 04:02:11,859][__main__][INFO] - agents played in iteration 352 are Alice, Bob [2026-03-26 04:02:12,759][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:02:12,803][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:02:12,804][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:02:12,804][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:02:13,413][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:02:13,869][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:02:14,373][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:02:14,872][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:02:15,372][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:02:15,870][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:02:16,368][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:02:16,867][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:02:17,366][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:02:17,865][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:02:18,367][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:02:18,866][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:02:19,365][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:02:19,864][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:02:20,364][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:02:20,862][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:02:21,362][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:02:21,863][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:02:22,364][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:02:22,864][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:02:23,365][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:02:23,866][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:02:24,366][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:02:24,867][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:02:25,368][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:02:25,868][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:02:26,367][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:02:26,865][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:02:27,365][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:02:27,865][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:02:28,365][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:02:28,864][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:02:29,363][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:02:29,864][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:02:30,364][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:02:30,864][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:02:31,365][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:02:31,866][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:02:32,366][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:02:32,866][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:02:33,365][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:02:33,865][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:02:34,364][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:02:34,863][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:02:35,362][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:02:35,861][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:02:36,360][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:02:36,859][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:02:37,359][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:02:37,859][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:02:38,359][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:02:38,859][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:02:39,357][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:02:39,856][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:02:40,355][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:02:40,853][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:02:41,352][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:02:41,850][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:02:42,349][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:02:42,847][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:02:43,347][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:02:43,845][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:02:44,345][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:02:44,844][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:02:45,344][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:02:45,847][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:02:46,348][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:02:46,847][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:02:47,346][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:02:47,846][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:02:48,347][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:02:48,847][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:02:49,346][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:02:49,846][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:02:50,345][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:02:50,844][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:02:51,342][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:02:51,842][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:02:52,342][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:02:52,841][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:02:53,340][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:02:53,840][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:02:54,339][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:02:54,838][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:02:55,339][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:02:55,841][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:02:56,343][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:02:56,844][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:02:57,344][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:02:57,843][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:02:58,342][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:02:58,842][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:02:59,341][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:02:59,839][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:03:00,339][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:03:00,838][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:03:01,337][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:03:01,837][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:03:02,337][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:03:02,837][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:03:03,337][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:03:03,838][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:03:04,338][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:03:04,838][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:03:05,339][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:03:05,838][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:03:06,338][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:03:06,836][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:03:07,337][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:03:07,836][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:03:08,336][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:03:08,835][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:03:09,333][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:03:09,831][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:03:10,329][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:03:10,826][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:03:11,325][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:03:11,823][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:03:12,322][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:03:12,820][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:03:13,318][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:03:13,815][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:03:14,314][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:03:14,812][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:03:15,309][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:03:15,807][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:03:16,305][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:03:16,804][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:03:17,302][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21706 tokens. [2026-03-26 04:03:17,935][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 04:03:18,676][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:03:18,679][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:03:18,680][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:03:19,415][__main__][INFO] - Iteration 353 took 1m 14s (9.10% Gen, 89.91% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 27m 32s. Estimated total time: 61h 56m 9s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 52s, 500 more iterations: 10h 19m 21s. [2026-03-26 04:03:19,417][__main__][INFO] - Starting iteration 353. [2026-03-26 04:03:19,820][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:03:19,821][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:03:20,411][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:03:20,416][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:03:26,591][__main__][INFO] - Number of regex retries in iteration 353: 2 [2026-03-26 04:03:26,592][__main__][INFO] - agents played in iteration 353 are Alice, Bob [2026-03-26 04:03:27,464][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:03:27,509][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:03:27,509][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:03:27,510][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:03:28,113][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:03:28,572][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:03:29,076][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:03:29,576][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:03:30,074][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:03:30,573][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:03:31,073][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:03:31,572][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:03:32,072][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:03:32,571][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:03:33,070][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:03:33,568][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:03:34,066][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:03:34,567][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:03:35,066][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:03:35,565][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:03:36,065][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:03:36,565][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:03:37,065][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:03:37,564][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:03:38,064][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:03:38,564][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:03:39,063][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:03:39,563][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:03:40,062][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:03:40,562][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:03:41,060][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:03:41,559][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:03:42,058][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:03:42,559][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:03:43,057][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:03:43,555][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:03:44,052][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:03:44,552][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:03:45,050][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:03:45,549][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:03:46,048][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:03:46,547][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:03:47,045][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:03:47,545][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:03:48,045][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:03:48,545][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:03:49,044][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:03:49,542][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:03:50,041][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:03:50,540][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:03:51,039][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:03:51,536][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:03:52,033][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:03:52,531][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:03:53,029][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:03:53,526][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:03:54,025][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:03:54,523][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:03:55,022][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:03:55,520][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:03:56,019][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:03:56,517][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:03:57,015][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:03:57,513][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:03:58,011][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:03:58,510][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:03:59,009][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:03:59,507][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:04:00,006][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:04:00,506][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:04:01,005][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:04:01,504][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:04:02,003][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:04:02,502][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:04:03,002][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:04:03,502][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:04:04,000][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:04:04,499][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:04:04,997][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:04:05,495][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:04:05,993][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:04:06,492][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:04:06,990][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:04:07,488][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:04:07,987][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:04:08,486][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:04:08,985][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:04:09,483][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:04:09,982][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:04:10,482][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:04:10,980][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:04:11,480][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:04:11,979][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:04:12,477][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:04:12,975][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:04:13,473][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:04:13,971][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:04:14,469][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:04:14,968][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:04:15,469][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:04:15,970][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:04:16,470][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:04:16,968][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:04:17,466][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:04:17,966][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:04:18,467][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:04:18,966][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:04:19,465][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:04:19,963][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:04:20,461][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:04:20,959][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:04:21,456][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:04:21,954][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:04:22,451][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:04:22,947][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:04:23,445][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:04:23,942][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:04:24,439][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:04:24,936][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:04:25,434][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:04:25,931][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:04:26,428][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:04:26,925][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:04:27,422][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:04:27,920][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:04:28,419][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:04:28,917][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:04:29,414][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:04:29,912][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:04:30,410][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:04:30,907][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:04:31,405][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:04:31,903][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 04:04:32,531][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 04:04:33,269][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:04:33,271][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:04:33,273][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:04:34,003][__main__][INFO] - Iteration 354 took 1m 14s (9.13% Gen, 89.89% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 19m 17s. Estimated total time: 61h 49m 9s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 38s, 500 more iterations: 10h 18m 11s. [2026-03-26 04:04:34,005][__main__][INFO] - Starting iteration 354. [2026-03-26 04:04:34,417][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:04:34,418][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:04:35,004][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:04:37,747][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:04:39,272][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:04:41,128][__main__][INFO] - Number of regex retries in iteration 354: 3 [2026-03-26 04:04:41,129][__main__][INFO] - agents played in iteration 354 are Alice, Bob [2026-03-26 04:04:42,008][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:04:42,053][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:04:42,053][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:04:42,054][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:04:42,670][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:04:43,125][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:04:43,629][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:04:44,128][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:04:44,626][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:04:45,126][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:04:45,624][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:04:46,123][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:04:46,622][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:04:47,120][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:04:47,618][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:04:48,116][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:04:48,616][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:04:49,113][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:04:49,610][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:04:50,109][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:04:50,608][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:04:51,110][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:04:51,613][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:04:52,114][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:04:52,616][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:04:53,119][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:04:53,622][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:04:54,123][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:04:54,625][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:04:55,125][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:04:55,625][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:04:56,127][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:04:56,626][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:04:57,126][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:04:57,624][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:04:58,123][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:04:58,622][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:04:59,121][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:04:59,621][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:05:00,120][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:05:00,618][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:05:01,117][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:05:01,615][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:05:02,112][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:05:02,611][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:05:03,111][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:05:03,610][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:05:04,109][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:05:04,608][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:05:05,108][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:05:05,607][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:05:06,106][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:05:06,606][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:05:07,106][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:05:07,605][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:05:08,105][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:05:08,604][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:05:09,103][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:05:09,602][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:05:10,102][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:05:10,601][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:05:11,101][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:05:11,601][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:05:12,101][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:05:12,600][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:05:13,100][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:05:13,600][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:05:14,098][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:05:14,600][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:05:15,100][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:05:15,600][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:05:16,099][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:05:16,597][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:05:17,096][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:05:17,595][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:05:18,094][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:05:18,595][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:05:19,095][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:05:19,594][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:05:20,093][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:05:20,592][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:05:21,091][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:05:21,590][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:05:22,089][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:05:22,589][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:05:23,090][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:05:23,593][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:05:24,094][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:05:24,596][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:05:25,100][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:05:25,603][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:05:26,105][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:05:26,607][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:05:27,106][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:05:27,605][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:05:28,104][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:05:28,604][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:05:29,103][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:05:29,602][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:05:30,100][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:05:30,599][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:05:31,097][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:05:31,596][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:05:32,093][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:05:32,591][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:05:33,089][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:05:33,586][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:05:34,085][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:05:34,583][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:05:35,083][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:05:35,584][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:05:36,084][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:05:36,583][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:05:37,083][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:05:37,584][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:05:38,084][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:05:38,584][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:05:39,083][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:05:39,581][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:05:40,081][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:05:40,580][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:05:41,079][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:05:41,578][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:05:42,077][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:05:42,576][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:05:43,075][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:05:43,575][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:05:44,074][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:05:44,574][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:05:45,074][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:05:45,577][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:05:46,080][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:05:46,580][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 04:05:47,227][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 04:05:47,968][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:05:47,970][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:05:47,972][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:05:48,624][__main__][INFO] - Iteration 355 took 1m 14s (9.04% Gen, 90.08% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 19m 16s. Estimated total time: 61h 50m 22s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 40s, 500 more iterations: 10h 18m 23s. [2026-03-26 04:05:48,626][__main__][INFO] - Starting iteration 355. [2026-03-26 04:05:49,032][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:05:49,032][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:05:55,361][__main__][INFO] - Number of regex retries in iteration 355: 0 [2026-03-26 04:05:55,362][__main__][INFO] - agents played in iteration 355 are Alice, Bob [2026-03-26 04:05:56,238][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:05:56,283][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:05:56,284][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:05:56,284][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:05:56,895][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:05:57,354][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:05:57,856][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:05:58,356][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:05:58,856][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:05:59,355][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:05:59,854][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:06:00,354][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:06:00,853][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:06:01,351][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:06:01,850][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:06:02,350][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:06:02,850][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:06:03,348][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:06:03,846][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:06:04,345][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:06:04,845][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:06:05,345][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:06:05,844][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:06:06,344][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:06:06,844][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:06:07,342][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:06:07,842][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:06:08,342][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:06:08,841][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:06:09,342][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:06:09,840][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:06:10,339][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:06:10,838][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:06:11,336][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:06:11,834][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:06:12,332][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:06:12,831][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:06:13,330][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:06:13,828][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:06:14,326][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:06:14,825][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:06:15,324][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:06:15,823][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:06:16,321][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:06:16,820][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:06:17,321][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:06:17,821][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:06:18,323][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:06:18,823][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:06:19,323][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:06:19,822][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:06:20,322][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:06:20,822][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:06:21,321][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:06:21,819][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:06:22,317][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:06:22,815][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:06:23,313][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:06:23,811][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:06:24,308][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:06:24,807][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:06:25,306][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:06:25,806][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:06:26,305][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:06:26,804][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:06:27,303][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:06:27,802][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:06:28,301][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:06:28,801][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:06:29,301][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:06:29,801][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:06:30,301][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:06:30,802][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:06:31,302][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:06:31,802][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:06:32,302][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:06:32,802][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:06:33,301][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:06:33,799][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:06:34,297][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:06:34,796][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:06:35,294][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:06:35,792][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:06:36,291][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:06:36,790][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:06:37,289][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:06:37,788][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:06:38,286][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:06:38,785][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:06:39,284][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:06:39,783][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:06:40,282][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:06:40,781][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:06:41,281][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:06:41,779][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:06:42,278][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:06:42,777][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:06:43,275][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:06:43,774][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:06:44,273][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:06:44,772][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:06:45,271][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:06:45,769][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:06:46,267][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:06:46,769][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:06:47,267][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:06:47,767][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:06:48,268][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:06:48,768][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:06:49,267][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:06:49,767][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:06:50,266][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:06:50,766][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:06:51,265][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:06:51,764][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:06:52,264][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:06:52,764][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:06:53,262][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:06:53,761][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:06:54,261][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:06:54,761][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:06:55,261][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:06:55,760][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:06:56,259][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:06:56,758][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:06:57,256][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:06:57,756][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:06:58,254][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:06:58,753][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:06:59,252][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:06:59,751][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:07:00,251][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:07:00,750][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 04:07:01,385][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 04:07:38,941][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:07:38,943][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:07:38,945][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:07:40,058][__main__][INFO] - Iteration 356 took 1m 51s (5.70% Gen, 93.30% Train). Generation: 6s, Training: 1m 43s. Estimated remaining time: 84h 58m 21s. Estimated total time: 92h 31m 19s. Time estimates for 10 more iterations: 18m 30s, 100 more iterations: 3h 5m 2s, 500 more iterations: 15h 25m 13s. [2026-03-26 04:07:40,062][__main__][INFO] - Starting iteration 356. [2026-03-26 04:07:40,461][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:07:40,462][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:07:41,059][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:07:42,658][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:07:47,222][__main__][INFO] - Number of regex retries in iteration 356: 2 [2026-03-26 04:07:47,223][__main__][INFO] - agents played in iteration 356 are Alice, Bob [2026-03-26 04:07:48,118][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:07:48,163][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:07:48,163][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:07:48,164][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:07:48,783][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:07:49,246][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:07:49,751][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:07:50,251][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:07:50,750][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:07:51,249][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:07:51,750][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:07:52,249][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:07:52,749][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:07:53,248][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:07:53,747][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:07:54,245][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:07:54,745][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:07:55,245][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:07:55,744][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:07:56,244][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:07:56,743][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:07:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:07:57,742][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:07:58,241][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:07:58,741][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:07:59,242][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:07:59,743][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:08:00,241][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:08:00,740][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:08:01,239][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:08:01,739][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:08:02,237][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:08:02,737][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:08:03,236][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:08:03,735][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:08:04,235][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:08:04,734][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:08:05,233][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:08:05,732][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:08:06,230][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:08:06,729][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:08:07,229][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:08:07,733][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:08:08,232][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:08:08,731][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:08:09,230][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:08:09,729][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:08:10,229][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:08:10,728][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:08:11,228][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:08:11,727][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:08:12,226][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:08:12,726][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:08:13,225][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:08:13,724][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:08:14,223][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:08:14,723][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:08:15,221][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:08:15,720][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:08:16,218][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:08:16,717][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:08:17,214][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:08:17,712][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:08:18,212][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:08:18,711][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:08:19,210][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:08:19,708][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:08:20,207][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:08:20,706][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:08:21,205][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:08:21,705][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:08:22,204][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:08:22,704][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:08:23,204][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:08:23,703][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:08:24,202][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:08:24,701][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:08:25,200][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:08:25,699][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:08:26,200][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:08:26,700][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:08:27,198][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:08:27,697][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:08:28,196][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:08:28,695][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:08:29,194][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:08:29,692][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:08:30,190][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:08:30,690][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:08:31,189][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:08:31,688][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:08:32,187][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:08:32,687][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:08:33,186][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:08:33,685][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:08:34,184][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:08:34,684][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:08:35,182][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:08:35,681][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:08:36,181][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:08:36,680][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:08:37,181][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:08:37,681][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:08:38,180][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:08:38,680][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:08:39,178][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:08:39,677][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:08:40,175][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:08:40,674][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:08:41,172][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:08:41,670][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:08:42,169][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:08:42,667][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:08:43,165][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:08:43,664][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:08:44,164][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:08:44,663][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:08:45,162][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:08:45,666][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:08:46,166][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:08:46,666][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:08:47,165][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:08:47,666][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:08:48,166][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:08:48,668][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:08:49,168][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:08:49,666][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:08:50,166][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:08:50,665][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:08:51,166][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:08:51,665][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:08:52,164][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:08:52,662][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 04:08:53,304][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 04:08:54,048][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:08:54,051][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:08:54,055][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:08:54,740][__main__][INFO] - Iteration 357 took 1m 14s (9.10% Gen, 89.98% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 19m 44s. Estimated total time: 61h 53m 57s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 59s. [2026-03-26 04:08:54,742][__main__][INFO] - Starting iteration 357. [2026-03-26 04:08:55,142][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:08:55,143][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:09:01,896][__main__][INFO] - Number of regex retries in iteration 357: 0 [2026-03-26 04:09:01,897][__main__][INFO] - agents played in iteration 357 are Alice, Bob [2026-03-26 04:09:02,776][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:09:02,821][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:09:02,822][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:09:02,822][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:09:03,430][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:09:03,884][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:09:04,386][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:09:04,884][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:09:05,382][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:09:05,881][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:09:06,380][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:09:06,879][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:09:07,377][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:09:07,877][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:09:08,377][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:09:08,876][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:09:09,376][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:09:09,876][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:09:10,376][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:09:10,876][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:09:11,377][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:09:11,875][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:09:12,373][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:09:12,871][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:09:13,370][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:09:13,869][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:09:14,367][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:09:14,866][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:09:15,364][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:09:15,866][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:09:16,367][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:09:16,869][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:09:17,370][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:09:17,872][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:09:18,375][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:09:18,875][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:09:19,376][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:09:19,876][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:09:20,375][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:09:20,875][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:09:21,376][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:09:21,876][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:09:22,375][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:09:22,874][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:09:23,374][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:09:23,873][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:09:24,371][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:09:24,869][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:09:25,368][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:09:25,867][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:09:26,365][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:09:26,865][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:09:27,365][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:09:27,864][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:09:28,363][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:09:28,862][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:09:29,362][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:09:29,861][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:09:30,360][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:09:30,859][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:09:31,358][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:09:31,856][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:09:32,355][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:09:32,854][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:09:33,352][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:09:33,850][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:09:34,348][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:09:34,847][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:09:35,345][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:09:35,845][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:09:36,344][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:09:36,842][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:09:37,342][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:09:37,840][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:09:38,338][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:09:38,836][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:09:39,333][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:09:39,833][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:09:40,331][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:09:40,830][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:09:41,330][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:09:41,829][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:09:42,328][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:09:42,827][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:09:43,327][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:09:43,824][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:09:44,322][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:09:44,820][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:09:45,318][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:09:45,816][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:09:46,313][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:09:46,810][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:09:47,308][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:09:47,809][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:09:48,312][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:09:48,812][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:09:49,313][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:09:49,814][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:09:50,313][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:09:50,813][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:09:51,314][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:09:51,814][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:09:52,313][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:09:52,813][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:09:53,313][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:09:53,813][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:09:54,312][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:09:54,811][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:09:55,313][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:09:55,811][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:09:56,310][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:09:56,808][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:09:57,307][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:09:57,806][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:09:58,304][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:09:58,803][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:09:59,303][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:09:59,802][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:10:00,307][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:10:00,805][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:10:01,304][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:10:01,803][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:10:02,302][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:10:02,801][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:10:03,302][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:10:03,801][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:10:04,299][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:10:04,797][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:10:05,295][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:10:05,793][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:10:06,291][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:10:06,790][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:10:07,289][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 04:10:07,931][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 04:10:08,732][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:10:08,735][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:10:08,736][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:10:09,558][__main__][INFO] - Iteration 358 took 1m 14s (9.08% Gen, 89.82% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 25m 21s. Estimated total time: 62h 0m 48s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 1s, 500 more iterations: 10h 20m 8s. [2026-03-26 04:10:09,560][__main__][INFO] - Starting iteration 358. [2026-03-26 04:10:09,968][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:10:09,968][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:10:16,918][__main__][INFO] - Number of regex retries in iteration 358: 0 [2026-03-26 04:10:16,919][__main__][INFO] - agents played in iteration 358 are Alice, Bob [2026-03-26 04:10:17,799][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:10:17,844][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:10:17,845][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:10:17,845][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:10:18,451][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:10:18,907][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:10:19,411][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:10:19,911][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:10:20,411][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:10:20,910][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:10:21,410][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:10:21,908][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:10:22,408][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:10:22,906][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:10:23,404][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:10:23,902][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:10:24,400][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:10:24,897][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:10:25,395][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:10:25,892][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:10:26,390][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:10:26,891][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:10:27,391][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:10:27,891][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:10:28,391][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:10:28,891][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:10:29,391][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:10:29,891][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:10:30,391][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:10:30,889][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:10:31,389][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:10:31,888][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:10:32,387][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:10:32,886][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:10:33,386][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:10:33,885][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:10:34,384][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:10:34,883][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:10:35,382][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:10:35,881][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:10:36,379][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:10:36,878][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:10:37,377][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:10:37,876][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:10:38,375][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:10:38,875][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:10:39,374][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:10:39,873][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:10:40,371][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:10:40,871][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:10:41,370][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:10:41,869][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:10:42,368][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:10:42,868][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:10:43,369][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:10:43,869][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:10:44,370][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:10:44,870][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:10:45,371][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:10:45,871][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:10:46,371][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:10:46,870][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:10:47,369][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:10:47,869][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:10:48,371][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:10:48,871][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:10:49,371][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:10:49,870][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:10:50,369][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:10:50,868][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:10:51,368][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:10:51,868][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:10:52,367][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:10:52,866][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:10:53,365][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:10:53,864][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:10:54,364][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:10:54,864][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:10:55,363][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:10:55,863][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:10:56,361][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:10:56,859][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:10:57,357][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:10:57,854][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:10:58,352][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:10:58,850][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:10:59,349][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:10:59,848][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:11:00,347][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:11:00,846][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:11:01,345][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:11:01,844][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:11:02,344][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:11:02,843][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:11:03,343][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:11:03,842][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:11:04,341][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:11:04,840][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:11:05,339][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:11:05,838][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:11:06,336][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:11:06,833][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:11:07,332][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:11:07,832][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:11:08,331][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:11:08,830][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:11:09,329][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:11:09,828][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:11:10,327][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:11:10,826][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:11:11,325][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:11:11,824][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:11:12,322][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:11:12,821][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:11:13,319][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:11:13,818][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:11:14,316][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:11:14,814][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:11:15,313][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:11:15,812][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:11:16,311][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:11:16,810][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:11:17,309][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:11:17,808][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:11:18,310][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:11:18,809][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:11:19,308][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:11:19,808][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:11:20,307][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:11:20,806][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:11:21,305][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:11:21,804][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:11:22,303][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 04:11:22,928][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:11:24,757][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:11:24,760][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:11:24,761][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:11:26,637][__main__][INFO] - Iteration 359 took 1m 16s (9.07% Gen, 88.49% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 56h 16m 46s. Estimated total time: 63h 53m 30s. Time estimates for 10 more iterations: 12m 46s, 100 more iterations: 2h 7m 47s, 500 more iterations: 10h 38m 55s. [2026-03-26 04:11:26,640][__main__][INFO] - Starting iteration 359. [2026-03-26 04:11:27,040][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:11:27,041][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:11:30,267][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:11:33,659][__main__][INFO] - Number of regex retries in iteration 359: 1 [2026-03-26 04:11:33,660][__main__][INFO] - agents played in iteration 359 are Alice, Bob [2026-03-26 04:11:34,549][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:11:34,593][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:11:34,594][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:11:34,595][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:11:35,197][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:11:35,652][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:11:36,155][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:11:36,654][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:11:37,152][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:11:37,651][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:11:38,150][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:11:38,649][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:11:39,148][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:11:39,650][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:11:40,152][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:11:40,652][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:11:41,152][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:11:41,654][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:11:42,155][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:11:42,656][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:11:43,156][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:11:43,655][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:11:44,155][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:11:44,654][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:11:45,154][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:11:45,656][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:11:46,156][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:11:46,655][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:11:47,153][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:11:47,652][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:11:48,152][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:11:48,652][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:11:49,151][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:11:49,649][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:11:50,149][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:11:50,648][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:11:51,146][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:11:51,646][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:11:52,146][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:11:52,645][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:11:53,144][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:11:53,643][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:11:54,144][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:11:54,643][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:11:55,144][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:11:55,644][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:11:56,145][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:11:56,644][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:11:57,147][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:11:57,647][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:11:58,147][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:11:58,647][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:11:59,148][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:11:59,647][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:12:00,145][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:12:00,644][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:12:01,143][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:12:01,643][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:12:02,144][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:12:02,643][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:12:03,143][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:12:03,643][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:12:04,143][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:12:04,644][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:12:05,146][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:12:05,646][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:12:06,146][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:12:06,645][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:12:07,144][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:12:07,641][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:12:08,139][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:12:08,636][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:12:09,133][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:12:09,631][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:12:10,130][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:12:10,628][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:12:11,126][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:12:11,625][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:12:12,125][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:12:12,623][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:12:13,123][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:12:13,623][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:12:14,123][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:12:14,623][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:12:15,123][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:12:15,621][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:12:16,120][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:12:16,619][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:12:17,118][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:12:17,617][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:12:18,117][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:12:18,617][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:12:19,116][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:12:19,614][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:12:20,112][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:12:20,610][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:12:21,109][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:12:21,608][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:12:22,106][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:12:22,604][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:12:23,104][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:12:23,603][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:12:24,102][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:12:24,600][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:12:25,101][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:12:25,603][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:12:26,101][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:12:26,601][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:12:27,100][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:12:27,599][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:12:28,098][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:12:28,597][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:12:29,097][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:12:29,597][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:12:30,097][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:12:30,596][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:12:31,096][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:12:31,595][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:12:32,094][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:12:32,592][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:12:33,091][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:12:33,591][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:12:34,090][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:12:34,589][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:12:35,087][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:12:35,586][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:12:36,086][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:12:36,585][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:12:37,085][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:12:37,585][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:12:38,084][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:12:38,583][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:12:39,082][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 04:12:39,710][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:12:40,587][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:12:40,590][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:12:40,592][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:12:41,296][__main__][INFO] - Iteration 360 took 1m 14s (8.91% Gen, 90.14% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 14m 49s. Estimated total time: 61h 52m 48s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 45s, 500 more iterations: 10h 18m 48s. [2026-03-26 04:12:41,299][__main__][INFO] - Starting iteration 360. [2026-03-26 04:12:41,704][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:12:41,704][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:12:48,112][__main__][INFO] - Number of regex retries in iteration 360: 0 [2026-03-26 04:12:48,114][__main__][INFO] - agents played in iteration 360 are Alice, Bob [2026-03-26 04:12:49,004][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:12:49,324][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:12:49,325][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:12:49,325][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:12:49,926][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:12:50,381][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:12:50,886][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:12:51,386][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:12:51,884][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:12:52,383][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:12:52,881][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:12:53,381][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:12:53,882][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:12:54,380][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:12:54,879][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:12:55,379][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:12:55,877][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:12:56,375][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:12:56,873][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:12:57,372][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:12:57,871][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:12:58,369][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:12:58,867][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:12:59,365][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:12:59,863][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:13:00,360][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:13:00,857][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:13:01,354][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:13:01,851][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:13:02,351][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:13:02,851][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:13:03,349][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:13:03,847][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:13:04,346][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:13:04,845][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:13:05,343][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:13:05,843][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:13:06,343][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:13:06,842][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:13:07,340][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:13:07,838][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:13:08,337][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:13:08,834][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:13:09,332][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:13:09,831][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:13:10,331][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:13:10,830][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:13:11,329][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:13:11,829][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:13:12,328][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:13:12,827][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:13:13,325][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:13:13,826][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:13:14,324][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:13:14,822][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:13:15,321][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:13:15,822][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:13:16,320][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:13:16,817][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:13:17,314][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:13:17,812][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:13:18,317][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:13:18,818][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:13:19,318][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:13:19,819][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:13:20,319][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:13:20,819][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:13:21,318][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:13:21,819][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:13:22,318][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:13:22,816][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:13:23,314][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:13:23,811][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:13:24,310][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:13:24,808][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:13:25,308][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:13:25,806][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:13:26,304][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:13:26,802][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:13:27,299][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:13:27,797][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:13:28,295][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:13:28,793][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:13:29,291][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:13:29,789][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:13:30,287][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:13:30,784][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:13:31,282][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:13:31,779][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:13:32,276][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:13:32,774][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:13:33,272][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:13:33,769][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:13:34,268][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:13:34,766][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:13:35,263][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:13:35,761][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:13:36,257][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:13:36,755][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:13:37,252][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:13:37,749][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:13:38,248][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:13:38,746][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:13:39,244][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:13:39,743][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:13:40,241][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:13:40,739][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:13:41,237][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:13:41,734][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:13:42,232][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:13:42,732][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:13:43,232][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:13:43,731][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:13:44,230][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:13:44,729][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:13:45,228][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:13:45,727][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:13:46,224][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:13:46,722][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:13:47,219][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:13:47,715][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:13:48,214][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:13:48,712][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:13:49,208][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:13:49,704][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:13:50,203][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:13:50,704][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:13:51,204][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:13:51,705][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:13:52,205][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:13:52,705][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:13:53,204][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:13:53,705][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 04:13:54,337][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.67%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 04:13:55,082][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:13:55,085][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:13:55,086][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:13:55,813][__main__][INFO] - Iteration 361 took 1m 14s (8.65% Gen, 90.37% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 6m 16s. Estimated total time: 61h 45m 30s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 31s, 500 more iterations: 10h 17m 35s. [2026-03-26 04:13:55,816][__main__][INFO] - Starting iteration 361. [2026-03-26 04:13:56,220][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:13:56,220][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:13:56,807][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:13:58,365][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:13:58,946][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:14:00,533][mllm.models.large_language_model_local][WARNING] - Response Given the high value of books for you and the high value of balls for Bob, a strategic proposal might involve splitting the books and balls to gain more points. Here’s a revised proposal: Proposal: 10 hats, 10 books, 10 balls Although this maintains a balanced approach, let's adjust it to potentially maximize your points: Proposal: 10 hats, 10 books, 5 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:14:03,204][__main__][INFO] - Number of regex retries in iteration 361: 4 [2026-03-26 04:14:03,205][__main__][INFO] - agents played in iteration 361 are Alice, Bob [2026-03-26 04:14:04,082][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:14:04,402][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:14:04,403][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:14:04,403][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:14:05,010][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:14:05,466][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:14:05,971][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:14:06,471][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:14:06,972][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:14:07,472][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:14:07,972][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:14:08,472][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:14:08,972][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:14:09,472][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:14:09,970][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:14:10,469][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:14:10,968][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:14:11,466][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:14:11,965][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:14:12,465][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:14:12,965][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:14:13,466][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:14:13,966][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:14:14,465][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:14:14,965][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:14:15,465][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:14:15,966][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:14:16,465][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:14:16,965][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:14:17,464][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:14:17,963][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:14:18,465][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:14:18,963][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:14:19,461][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:14:19,959][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:14:20,457][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:14:20,955][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:14:21,456][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:14:21,957][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:14:22,457][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:14:22,957][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:14:23,457][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:14:23,956][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:14:24,456][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:14:24,956][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:14:25,456][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:14:25,955][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:14:26,452][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:14:26,951][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:14:27,450][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:14:27,949][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:14:28,448][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:14:28,946][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:14:29,445][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:14:29,944][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:14:30,441][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:14:30,939][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:14:31,437][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:14:31,934][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:14:32,431][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:14:32,930][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:14:33,429][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:14:33,927][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:14:34,426][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:14:34,925][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:14:35,423][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:14:35,921][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:14:36,421][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:14:36,920][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:14:37,420][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:14:37,919][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:14:38,419][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:14:38,919][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:14:39,418][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:14:39,917][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:14:40,416][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:14:40,915][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:14:41,413][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:14:41,912][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:14:42,410][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:14:42,909][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:14:43,407][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:14:43,906][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:14:44,406][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:14:44,905][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:14:45,407][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:14:45,909][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:14:46,409][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:14:46,909][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:14:47,408][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:14:47,908][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:14:48,409][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:14:48,909][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:14:49,407][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:14:49,905][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:14:50,402][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:14:50,899][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:14:51,397][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:14:51,894][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:14:52,392][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:14:52,890][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:14:53,391][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:14:53,891][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:14:54,391][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:14:54,891][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:14:55,393][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:14:55,895][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:14:56,397][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:14:56,898][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:14:57,397][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:14:57,895][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:14:58,394][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:14:58,893][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:14:59,392][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:14:59,890][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:15:00,390][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:15:01,371][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:15:01,869][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:15:02,367][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:15:02,866][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:15:03,366][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:15:03,866][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:15:04,365][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:15:04,998][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:15:05,499][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:15:05,999][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:15:06,498][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:15:06,996][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:15:07,495][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:15:07,994][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:15:08,493][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:15:08,992][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:15:09,492][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 04:15:10,175][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:05 [2026-03-26 04:15:10,820][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:15:10,824][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:15:10,826][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:15:11,643][__main__][INFO] - Iteration 362 took 1m 15s (9.26% Gen, 89.65% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 55h 10m 41s. Estimated total time: 62h 51m 11s. Time estimates for 10 more iterations: 12m 34s, 100 more iterations: 2h 5m 42s, 500 more iterations: 10h 28m 31s. [2026-03-26 04:15:11,646][__main__][INFO] - Starting iteration 362. [2026-03-26 04:15:12,046][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:15:12,047][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:15:12,742][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:15:18,619][__main__][INFO] - Number of regex retries in iteration 362: 1 [2026-03-26 04:15:18,620][__main__][INFO] - agents played in iteration 362 are Alice, Bob [2026-03-26 04:15:19,534][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:15:19,579][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:15:19,579][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:15:19,580][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:15:20,182][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:15:20,639][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:15:21,143][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:15:21,644][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:15:22,144][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:15:22,643][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:15:23,143][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:15:23,642][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:15:24,141][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:15:24,639][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:15:25,140][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:15:25,639][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:15:26,138][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:15:26,637][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:15:27,136][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:15:27,636][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:15:28,134][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:15:28,633][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:15:29,133][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:15:29,633][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:15:30,132][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:15:30,631][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:15:31,130][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:15:31,630][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:15:32,129][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:15:32,627][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:15:33,126][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:15:33,625][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:15:34,124][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:15:34,623][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:15:35,121][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:15:35,619][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:15:36,117][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:15:36,618][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:15:37,119][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:15:37,619][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:15:38,120][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:15:38,619][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:15:39,120][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:15:39,621][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:15:40,121][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:15:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:15:41,120][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:15:41,619][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:15:42,118][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:15:42,617][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:15:43,116][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:15:43,615][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:15:44,115][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:15:44,615][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:15:45,114][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:15:45,614][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:15:46,113][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:15:46,613][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:15:47,113][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:15:47,613][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:15:48,113][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:15:48,615][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:15:49,114][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:15:49,613][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:15:50,112][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:15:50,611][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:15:51,110][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:15:51,610][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:15:52,109][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:15:52,608][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:15:53,107][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:15:53,607][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:15:54,106][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:15:54,605][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:15:55,106][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:15:55,605][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:15:56,106][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:15:56,606][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:15:57,104][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:15:57,603][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:15:58,102][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:15:58,602][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:15:59,101][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:15:59,600][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:16:00,100][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:16:00,598][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:16:01,097][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:16:01,596][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:16:02,095][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:16:02,594][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:16:03,093][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:16:03,592][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:16:04,092][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:16:04,590][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:16:05,088][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:16:05,586][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:16:06,086][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:16:06,586][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:16:07,085][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:16:07,584][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:16:08,083][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:16:08,583][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:16:09,083][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:16:09,582][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:16:10,084][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:16:10,585][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:16:11,086][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:16:11,586][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:16:12,087][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:16:12,586][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:16:13,085][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:16:13,586][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:16:14,086][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:16:14,586][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:16:15,085][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:16:15,587][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:16:16,089][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:16:16,589][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:16:17,088][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:16:17,587][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:16:18,088][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:16:18,589][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:16:19,089][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:16:19,589][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:16:20,089][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:16:20,588][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:16:21,088][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:16:21,589][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:16:22,089][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:16:22,589][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:16:23,089][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:16:23,589][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:16:24,090][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 04:16:24,778][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:16:25,531][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:16:25,534][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:16:25,535][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:16:26,321][__main__][INFO] - Iteration 363 took 1m 14s (8.85% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 12m 3s. Estimated total time: 61h 53m 47s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 57s. [2026-03-26 04:16:26,324][__main__][INFO] - Starting iteration 363. [2026-03-26 04:16:26,727][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:16:26,728][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:16:27,305][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:16:33,516][__main__][INFO] - Number of regex retries in iteration 363: 1 [2026-03-26 04:16:33,517][__main__][INFO] - agents played in iteration 363 are Alice, Bob [2026-03-26 04:16:34,396][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:16:34,440][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:16:34,441][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:16:34,441][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:16:35,052][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:16:35,509][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:16:36,013][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:16:36,514][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:16:37,014][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:16:37,513][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:16:38,014][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:16:38,514][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:16:39,013][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:16:39,512][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:16:40,011][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:16:40,509][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:16:41,008][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:16:41,507][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:16:42,007][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:16:42,506][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:16:43,005][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:16:43,505][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:16:44,005][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:16:44,505][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:16:45,005][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:16:45,506][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:16:46,007][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:16:46,506][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:16:47,006][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:16:47,507][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:16:48,009][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:16:48,511][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:16:49,012][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:16:49,512][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:16:50,012][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:16:50,513][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:16:51,013][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:16:51,512][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:16:52,011][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:16:52,510][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:16:53,009][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:16:53,507][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:16:54,006][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:16:54,504][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:16:55,002][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:16:55,500][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:16:55,999][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:16:56,497][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:16:56,995][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:16:57,494][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:16:57,993][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:16:58,492][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:16:58,990][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:16:59,489][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:16:59,989][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:17:00,488][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:17:00,987][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:17:01,489][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:17:01,990][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:17:02,488][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:17:02,988][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:17:03,486][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:17:03,986][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:17:04,486][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:17:04,985][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:17:05,484][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:17:05,982][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:17:06,480][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:17:06,978][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:17:07,478][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:17:07,978][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:17:08,478][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:17:08,977][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:17:09,477][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:17:09,976][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:17:10,475][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:17:10,975][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:17:11,473][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:17:11,972][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:17:12,471][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:17:12,969][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:17:13,467][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:17:13,966][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:17:14,465][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:17:14,963][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:17:15,463][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:17:15,965][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:17:16,465][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:17:16,964][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:17:17,463][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:17:17,963][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:17:18,465][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:17:18,966][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:17:19,466][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:17:19,967][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:17:20,467][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:17:20,967][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:17:21,468][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:17:21,967][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:17:22,468][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:17:22,969][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:17:23,468][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:17:23,967][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:17:24,465][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:17:24,964][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:17:25,462][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:17:25,960][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:17:26,459][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:17:26,957][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:17:27,457][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:17:27,955][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:17:28,453][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:17:28,951][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:17:29,450][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:17:29,949][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:17:30,448][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:17:30,947][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:17:31,446][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:17:31,945][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:17:32,443][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:17:32,941][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:17:33,439][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:17:33,937][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:17:34,436][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:17:34,934][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:17:35,433][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:17:35,931][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:17:36,429][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:17:36,928][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:17:37,426][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:17:37,924][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:17:38,423][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:17:38,923][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 04:17:39,558][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 04:17:40,331][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:17:40,334][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:17:40,335][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:17:41,107][__main__][INFO] - Iteration 364 took 1m 14s (9.13% Gen, 89.83% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 16m 4s. Estimated total time: 61h 59m 3s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 58s, 500 more iterations: 10h 19m 50s. [2026-03-26 04:17:41,109][__main__][INFO] - Starting iteration 364. [2026-03-26 04:17:41,511][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:17:41,512][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:17:42,098][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:17:43,795][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:17:48,018][__main__][INFO] - Number of regex retries in iteration 364: 2 [2026-03-26 04:17:48,019][__main__][INFO] - agents played in iteration 364 are Alice, Bob [2026-03-26 04:17:48,936][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:17:48,981][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:17:48,982][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:17:48,982][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:17:49,586][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:17:50,042][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:17:50,546][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:17:51,045][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:17:51,544][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:17:52,042][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:17:52,541][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:17:53,039][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:17:53,538][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:17:54,037][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:17:54,536][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:17:55,036][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:17:55,535][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:17:56,034][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:17:56,533][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:17:57,032][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:17:57,531][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:17:58,033][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:17:58,534][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:17:59,034][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:17:59,533][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:18:00,033][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:18:00,532][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:18:01,031][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:18:01,531][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:18:02,031][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:18:02,530][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:18:03,029][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:18:03,528][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:18:04,028][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:18:04,528][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:18:05,028][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:18:05,528][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:18:06,028][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:18:06,529][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:18:07,029][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:18:07,529][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:18:08,029][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:18:08,530][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:18:09,030][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:18:09,530][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:18:10,028][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:18:10,527][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:18:11,024][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:18:11,523][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:18:12,022][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:18:12,520][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:18:13,018][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:18:13,516][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:18:14,013][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:18:14,511][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:18:15,007][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:18:15,505][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:18:16,002][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:18:16,499][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:18:16,996][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:18:17,493][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:18:17,992][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:18:18,490][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:18:18,989][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:18:19,489][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:18:19,988][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:18:20,486][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:18:20,985][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:18:21,484][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:18:21,984][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:18:22,483][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:18:22,982][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:18:23,482][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:18:23,980][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:18:24,480][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:18:24,979][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:18:25,478][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:18:25,977][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:18:26,475][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:18:26,973][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:18:27,472][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:18:27,972][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:18:28,471][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:18:28,969][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:18:29,468][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:18:29,969][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:18:30,468][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:18:30,969][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:18:31,470][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:18:31,970][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:18:32,469][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:18:32,969][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:18:33,470][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:18:33,969][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:18:34,469][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:18:34,969][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:18:35,469][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:18:35,969][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:18:36,469][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:18:36,969][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:18:37,468][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:18:37,968][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:18:38,469][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:18:38,969][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:18:39,469][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:18:39,969][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:18:40,469][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:18:40,969][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:18:41,469][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:18:41,969][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:18:42,467][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:18:42,967][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:18:43,466][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:18:43,965][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:18:44,464][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:18:44,963][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:18:45,461][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:18:45,959][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:18:46,456][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:18:46,954][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:18:47,453][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:18:47,952][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:18:48,452][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:18:48,950][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:18:49,449][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:18:49,948][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:18:50,448][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:18:50,947][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:18:51,447][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:18:51,947][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:18:52,445][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:18:52,943][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:18:53,441][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:18:54,070][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 04:18:54,858][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:18:54,860][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:18:54,862][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:18:55,574][__main__][INFO] - Iteration 365 took 1m 14s (8.79% Gen, 90.25% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 58m 57s. Estimated total time: 61h 43m 11s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 26s, 500 more iterations: 10h 17m 11s. [2026-03-26 04:18:55,577][__main__][INFO] - Starting iteration 365. [2026-03-26 04:18:55,981][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:18:55,982][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:19:02,119][__main__][INFO] - Number of regex retries in iteration 365: 0 [2026-03-26 04:19:02,119][__main__][INFO] - agents played in iteration 365 are Alice, Bob [2026-03-26 04:19:02,995][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:19:03,040][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:19:03,040][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:19:03,041][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:19:03,648][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:19:04,103][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:19:04,608][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:19:05,107][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:19:05,606][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:19:06,104][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:19:06,602][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:19:07,100][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:19:07,599][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:19:08,097][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:19:08,594][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:19:09,092][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:19:09,590][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:19:10,089][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:19:10,588][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:19:11,086][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:19:11,583][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:19:12,082][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:19:12,582][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:19:13,080][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:19:13,580][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:19:14,078][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:19:14,577][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:19:15,075][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:19:15,577][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:19:16,081][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:19:16,581][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:19:17,080][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:19:17,580][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:19:18,080][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:19:18,581][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:19:19,081][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:19:19,581][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:19:20,080][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:19:20,580][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:19:21,080][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:19:21,579][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:19:22,078][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:19:22,577][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:19:23,077][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:19:23,576][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:19:24,079][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:19:24,580][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:19:25,080][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:19:25,580][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:19:26,080][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:19:26,581][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:19:27,080][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:19:27,581][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:19:28,080][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:19:28,579][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:19:29,077][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:19:29,576][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:19:30,075][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:19:30,574][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:19:31,073][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:19:31,572][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:19:32,072][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:19:32,571][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:19:33,070][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:19:33,570][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:19:34,069][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:19:34,568][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:19:35,067][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:19:35,566][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:19:36,065][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:19:36,563][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:19:37,062][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:19:37,561][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:19:38,060][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:19:38,558][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:19:39,057][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:19:39,555][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:19:40,052][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:19:40,550][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:19:41,048][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:19:41,546][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:19:42,046][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:19:42,546][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:19:43,043][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:19:43,541][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:19:44,038][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:19:44,537][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:19:45,035][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:19:45,533][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:19:46,032][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:19:46,530][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:19:47,028][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:19:47,527][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:19:48,027][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:19:48,529][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:19:49,027][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:19:49,528][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:19:50,027][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:19:50,527][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:19:51,026][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:19:51,525][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:19:52,025][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:19:52,523][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:19:53,021][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:19:53,520][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:19:54,018][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:19:54,517][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:19:55,017][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:19:55,516][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:19:56,016][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:19:56,516][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:19:57,015][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:19:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:19:58,015][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:19:58,515][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:19:59,015][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:19:59,514][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:20:00,013][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:20:00,511][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:20:01,009][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:20:01,508][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:20:02,007][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:20:02,506][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:20:03,004][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:20:03,503][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:20:04,001][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:20:04,499][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:20:04,997][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:20:05,496][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:20:05,996][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:20:06,495][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:20:06,994][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:20:07,493][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21760 tokens. [2026-03-26 04:20:08,135][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:20:08,821][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:20:08,823][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:20:08,825][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:20:09,600][__main__][INFO] - Iteration 366 took 1m 13s (8.34% Gen, 90.61% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 35m 30s. Estimated total time: 61h 20m 58s. Time estimates for 10 more iterations: 12m 16s, 100 more iterations: 2h 2m 41s, 500 more iterations: 10h 13m 29s. [2026-03-26 04:20:09,602][__main__][INFO] - Starting iteration 366. [2026-03-26 04:20:10,004][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:20:10,004][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:20:15,019][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:20:16,286][__main__][INFO] - Number of regex retries in iteration 366: 1 [2026-03-26 04:20:16,287][__main__][INFO] - agents played in iteration 366 are Alice, Bob [2026-03-26 04:20:17,172][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:20:17,216][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:20:17,217][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:20:17,218][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:20:17,829][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:20:18,284][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:20:18,788][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:20:19,288][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:20:19,787][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:20:20,286][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:20:20,784][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:20:21,282][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:20:21,781][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:20:22,278][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:20:22,776][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:20:23,275][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:20:23,775][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:20:24,274][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:20:24,772][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:20:25,271][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:20:25,770][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:20:26,271][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:20:26,771][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:20:27,271][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:20:27,771][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:20:28,270][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:20:28,769][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:20:29,269][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:20:29,769][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:20:30,268][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:20:30,767][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:20:31,266][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:20:31,764][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:20:32,263][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:20:32,760][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:20:33,258][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:20:33,757][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:20:34,253][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:20:34,748][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:20:35,244][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:20:35,739][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:20:36,234][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:20:36,729][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:20:37,224][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:20:37,720][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:20:38,218][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:20:38,718][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:20:39,217][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:20:39,715][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:20:40,214][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:20:40,712][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:20:41,211][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:20:41,710][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:20:42,213][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:20:42,716][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:20:43,218][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:20:43,719][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:20:44,221][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:20:44,723][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:20:45,225][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:20:45,731][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:20:46,231][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:20:46,729][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:20:47,228][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:20:47,726][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:20:48,226][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:20:48,725][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:20:49,223][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:20:49,720][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:20:50,217][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:20:50,716][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:20:51,214][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:20:51,713][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:20:52,212][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:20:52,710][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:20:53,209][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:20:53,708][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:20:54,206][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:20:54,706][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:20:55,205][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:20:55,704][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:20:56,202][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:20:56,701][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:20:57,199][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:20:57,697][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:20:58,198][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:20:58,696][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:20:59,196][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:20:59,695][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:21:00,195][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:21:00,694][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:21:01,193][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:21:01,694][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:21:02,194][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:21:02,694][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:21:03,192][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:21:03,691][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:21:04,190][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:21:04,688][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:21:05,186][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:21:05,683][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:21:06,178][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:21:06,673][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:21:07,168][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:21:07,663][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:21:08,158][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:21:08,654][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:21:09,149][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:21:09,645][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:21:10,143][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:21:10,641][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:21:11,140][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:21:11,639][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:21:12,139][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:21:12,637][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:21:13,136][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:21:13,634][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:21:14,136][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:21:14,638][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:21:15,139][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:21:15,640][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:21:16,142][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:21:16,644][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:21:17,145][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:21:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:21:18,149][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:21:18,649][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:21:19,147][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:21:19,645][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:21:20,143][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:21:20,641][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:21:21,138][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:21:21,636][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 04:21:22,277][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 04:21:23,825][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:21:23,827][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:21:23,829][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:21:24,794][__main__][INFO] - Iteration 367 took 1m 14s (8.40% Gen, 90.31% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 54h 32m 48s. Estimated total time: 62h 19m 31s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 39s, 500 more iterations: 10h 23m 15s. [2026-03-26 04:21:24,796][__main__][INFO] - Starting iteration 367. [2026-03-26 04:21:25,199][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:21:25,200][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:21:29,485][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:21:31,674][__main__][INFO] - Number of regex retries in iteration 367: 1 [2026-03-26 04:21:31,675][__main__][INFO] - agents played in iteration 367 are Alice, Bob [2026-03-26 04:21:32,565][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:21:32,610][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:21:32,610][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:21:32,611][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:21:33,222][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:21:33,677][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:21:34,181][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:21:34,680][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:21:35,179][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:21:35,678][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:21:36,178][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:21:36,676][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:21:37,175][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:21:37,676][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:21:38,178][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:21:38,678][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:21:39,179][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:21:39,680][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:21:40,180][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:21:40,680][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:21:41,179][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:21:41,679][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:21:42,178][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:21:42,677][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:21:43,176][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:21:43,675][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:21:44,174][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:21:44,674][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:21:45,174][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:21:45,674][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:21:46,174][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:21:46,673][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:21:47,173][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:21:47,673][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:21:48,175][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:21:48,675][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:21:49,174][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:21:49,676][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:21:50,177][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:21:50,677][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:21:51,177][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:21:51,677][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:21:52,177][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:21:52,676][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:21:53,177][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:21:53,675][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:21:54,175][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:21:54,674][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:21:55,173][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:21:55,673][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:21:56,172][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:21:56,672][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:21:57,171][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:21:57,668][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:21:58,167][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:21:58,664][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:21:59,161][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:21:59,658][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:22:00,156][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:22:00,655][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:22:01,152][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:22:01,651][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:22:02,151][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:22:02,650][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:22:03,150][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:22:03,648][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:22:04,149][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:22:04,648][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:22:05,147][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:22:05,646][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:22:06,145][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:22:06,643][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:22:07,141][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:22:07,639][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:22:08,137][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:22:08,635][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:22:09,134][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:22:09,633][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:22:10,133][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:22:10,633][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:22:11,133][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:22:11,634][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:22:12,135][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:22:12,635][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:22:13,135][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:22:13,634][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:22:14,133][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:22:14,632][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:22:15,132][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:22:15,635][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:22:16,136][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:22:16,637][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:22:17,136][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:22:17,635][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:22:18,134][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:22:18,635][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:22:19,135][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:22:19,634][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:22:20,133][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:22:20,632][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:22:21,131][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:22:21,631][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:22:22,132][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:22:22,631][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:22:23,132][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:22:23,631][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:22:24,131][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:22:24,631][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:22:25,132][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:22:25,631][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:22:26,130][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:22:26,629][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:22:27,127][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:22:27,626][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:22:28,124][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:22:28,624][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:22:29,121][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:22:29,619][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:22:30,117][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:22:30,615][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:22:31,114][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:22:31,613][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:22:32,111][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:22:32,609][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:22:33,107][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:22:33,606][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:22:34,105][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:22:34,604][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:22:35,103][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:22:35,602][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:22:36,101][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:22:36,600][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:22:37,099][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 04:22:37,736][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:22:38,536][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:22:38,538][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:22:38,540][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:22:39,685][__main__][INFO] - Iteration 368 took 1m 14s (8.69% Gen, 89.77% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 16m 23s. Estimated total time: 62h 4m 20s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 8s, 500 more iterations: 10h 20m 43s. [2026-03-26 04:22:39,688][__main__][INFO] - Starting iteration 368. [2026-03-26 04:22:40,090][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:22:40,091][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:22:46,431][__main__][INFO] - Number of regex retries in iteration 368: 0 [2026-03-26 04:22:46,432][__main__][INFO] - agents played in iteration 368 are Alice, Bob [2026-03-26 04:22:47,307][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:22:47,351][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:22:47,352][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:22:47,352][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:22:47,956][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:22:48,411][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:22:48,914][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:22:49,413][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:22:49,912][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:22:50,411][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:22:50,910][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:22:51,410][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:22:51,909][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:22:52,407][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:22:52,906][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:22:53,404][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:22:53,902][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:22:54,400][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:22:54,898][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:22:55,397][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:22:55,895][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:22:56,393][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:22:56,893][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:22:57,391][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:22:57,891][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:22:58,389][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:22:58,889][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:22:59,389][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:22:59,888][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:23:00,387][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:23:00,886][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:23:01,385][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:23:01,883][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:23:02,382][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:23:02,880][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:23:03,379][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:23:03,878][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:23:04,380][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:23:04,880][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:23:05,379][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:23:05,878][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:23:06,378][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:23:06,877][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:23:07,377][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:23:07,876][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:23:08,373][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:23:08,869][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:23:09,366][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:23:09,862][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:23:10,358][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:23:10,855][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:23:11,352][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:23:11,850][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:23:12,348][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:23:12,847][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:23:13,347][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:23:13,846][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:23:14,343][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:23:14,841][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:23:15,338][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:23:15,836][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:23:16,333][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:23:16,831][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:23:17,329][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:23:17,827][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:23:18,327][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:23:18,825][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:23:19,323][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:23:19,820][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:23:20,318][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:23:20,815][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:23:21,314][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:23:21,813][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:23:22,311][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:23:22,809][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:23:23,307][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:23:23,805][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:23:24,303][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:23:24,801][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:23:25,300][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:23:25,798][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:23:26,297][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:23:26,795][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:23:27,294][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:23:27,793][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:23:28,292][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:23:28,791][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:23:29,291][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:23:29,791][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:23:30,289][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:23:30,787][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:23:31,285][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:23:31,784][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:23:32,282][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:23:32,780][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:23:33,280][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:23:33,779][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:23:34,277][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:23:34,776][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:23:35,275][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:23:35,775][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:23:36,274][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:23:36,775][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:23:37,274][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:23:37,774][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:23:38,274][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:23:38,774][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:23:39,274][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:23:39,774][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:23:40,271][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:23:40,768][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:23:41,265][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:23:41,762][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:23:42,258][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:23:42,755][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:23:43,253][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:23:43,750][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:23:44,251][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:23:44,750][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:23:45,249][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:23:45,751][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:23:46,251][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:23:46,750][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:23:47,248][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:23:47,747][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:23:48,249][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:23:48,748][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:23:49,246][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:23:49,743][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:23:50,241][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:23:50,738][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:23:51,236][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:23:51,734][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:23:52,363][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 04:23:53,233][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:23:53,236][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:23:53,237][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:23:54,024][__main__][INFO] - Iteration 369 took 1m 13s (8.58% Gen, 90.36% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 47m 32s. Estimated total time: 61h 36m 44s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 13s, 500 more iterations: 10h 16m 7s. [2026-03-26 04:23:54,027][__main__][INFO] - Starting iteration 369. [2026-03-26 04:23:54,427][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:23:54,428][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:24:00,770][__main__][INFO] - Number of regex retries in iteration 369: 0 [2026-03-26 04:24:00,771][__main__][INFO] - agents played in iteration 369 are Alice, Bob [2026-03-26 04:24:01,649][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:24:01,693][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:24:01,694][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:24:01,694][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:24:02,301][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:24:02,756][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:24:03,259][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:24:03,758][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:24:04,257][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:24:04,757][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:24:05,256][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:24:05,756][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:24:06,255][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:24:06,755][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:24:07,253][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:24:07,753][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:24:08,252][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:24:08,751][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:24:09,251][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:24:09,750][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:24:10,248][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:24:10,746][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:24:11,243][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:24:11,740][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:24:12,237][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:24:12,735][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:24:13,235][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:24:13,733][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:24:14,231][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:24:14,730][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:24:15,229][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:24:15,728][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:24:16,226][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:24:16,726][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:24:17,225][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:24:17,723][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:24:18,224][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:24:18,722][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:24:19,219][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:24:19,716][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:24:20,214][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:24:20,711][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:24:21,209][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:24:21,707][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:24:22,205][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:24:22,703][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:24:23,201][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:24:23,699][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:24:24,197][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:24:24,696][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:24:25,196][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:24:25,695][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:24:26,193][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:24:26,693][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:24:27,192][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:24:27,691][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:24:28,190][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:24:28,689][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:24:29,189][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:24:29,689][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:24:30,189][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:24:30,687][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:24:31,185][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:24:31,684][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:24:32,181][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:24:32,679][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:24:33,177][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:24:33,677][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:24:34,175][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:24:34,674][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:24:35,173][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:24:35,673][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:24:36,173][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:24:36,672][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:24:37,172][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:24:37,671][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:24:38,171][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:24:38,671][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:24:39,169][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:24:39,668][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:24:40,169][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:24:40,668][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:24:41,167][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:24:41,667][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:24:42,166][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:24:42,664][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:24:43,162][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:24:43,660][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:24:44,159][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:24:44,657][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:24:45,157][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:24:45,655][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:24:46,154][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:24:46,653][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:24:47,151][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:24:47,651][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:24:48,150][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:24:48,652][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:24:49,151][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:24:49,650][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:24:50,150][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:24:50,647][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:24:51,144][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:24:51,641][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:24:52,138][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:24:52,635][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:24:53,133][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:24:53,631][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:24:54,129][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:24:54,628][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:24:55,126][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:24:55,625][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:24:56,123][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:24:56,620][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:24:57,118][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:24:57,617][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:24:58,116][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:24:58,614][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:24:59,113][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:24:59,612][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:25:00,111][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:25:00,611][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:25:01,110][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:25:01,610][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:25:02,109][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:25:02,609][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:25:03,106][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:25:03,604][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:25:04,101][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:25:04,600][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:25:05,098][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:25:05,596][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:25:06,094][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 04:25:06,735][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 04:25:07,525][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:25:07,528][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:25:07,530][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:25:08,196][__main__][INFO] - Iteration 370 took 1m 13s (8.60% Gen, 90.50% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 38m 0s. Estimated total time: 61h 28m 26s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 56s, 500 more iterations: 10h 14m 44s. [2026-03-26 04:25:08,198][__main__][INFO] - Starting iteration 370. [2026-03-26 04:25:08,595][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:25:08,596][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:25:09,194][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:25:16,046][__main__][INFO] - Number of regex retries in iteration 370: 1 [2026-03-26 04:25:16,047][__main__][INFO] - agents played in iteration 370 are Alice, Bob [2026-03-26 04:25:16,922][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:25:16,967][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:25:16,967][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:25:16,968][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:25:17,576][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:25:18,032][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:25:18,538][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:25:19,037][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:25:19,537][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:25:20,035][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:25:20,534][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:25:21,033][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:25:21,533][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:25:22,033][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:25:22,533][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:25:23,033][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:25:23,533][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:25:24,032][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:25:24,532][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:25:25,032][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:25:25,533][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:25:26,032][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:25:26,531][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:25:27,029][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:25:27,528][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:25:28,026][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:25:28,524][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:25:29,022][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:25:29,521][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:25:30,020][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:25:30,517][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:25:31,015][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:25:31,514][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:25:32,012][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:25:32,510][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:25:33,009][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:25:33,507][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:25:34,004][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:25:34,503][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:25:35,001][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:25:35,499][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:25:35,998][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:25:36,496][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:25:36,995][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:25:37,495][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:25:37,995][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:25:38,494][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:25:38,992][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:25:39,491][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:25:39,990][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:25:40,488][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:25:40,986][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:25:41,484][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:25:41,986][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:25:42,487][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:25:42,987][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:25:43,487][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:25:43,987][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:25:44,487][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:25:44,985][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:25:45,486][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:25:45,985][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:25:46,486][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:25:46,985][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:25:47,485][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:25:47,985][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:25:48,486][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:25:48,986][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:25:49,487][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:25:49,986][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:25:50,485][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:25:50,988][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:25:51,487][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:25:51,986][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:25:52,484][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:25:52,983][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:25:53,482][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:25:53,982][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:25:54,481][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:25:54,980][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:25:55,480][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:25:55,979][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:25:56,479][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:25:56,979][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:25:57,478][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:25:57,977][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:25:58,476][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:25:58,974][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:25:59,474][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:25:59,973][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:26:00,472][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:26:00,972][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:26:01,470][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:26:01,969][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:26:02,467][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:26:02,965][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:26:03,463][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:26:03,961][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:26:04,458][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:26:04,957][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:26:05,457][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:26:05,955][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:26:06,453][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:26:06,952][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:26:07,451][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:26:07,949][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:26:08,449][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:26:08,948][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:26:09,446][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:26:09,944][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:26:10,442][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:26:10,940][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:26:11,438][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:26:11,936][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:26:12,434][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:26:12,933][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:26:13,432][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:26:13,931][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:26:14,430][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:26:14,929][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:26:15,429][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:26:15,929][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:26:16,429][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:26:16,930][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:26:17,429][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:26:17,928][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:26:18,429][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:26:18,929][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:26:19,427][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:26:19,927][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:26:20,426][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:26:20,926][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:26:21,425][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 04:26:22,085][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 04:26:22,824][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:26:22,826][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:26:22,828][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:26:23,557][__main__][INFO] - Iteration 371 took 1m 14s (9.94% Gen, 89.09% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 54h 36m 25s. Estimated total time: 62h 28m 7s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 56s, 500 more iterations: 10h 24m 41s. [2026-03-26 04:26:23,560][__main__][INFO] - Starting iteration 371. [2026-03-26 04:26:23,959][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:26:23,960][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:26:26,275][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:26:31,079][__main__][INFO] - Number of regex retries in iteration 371: 1 [2026-03-26 04:26:31,079][__main__][INFO] - agents played in iteration 371 are Alice, Bob [2026-03-26 04:26:31,954][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:26:31,998][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:26:31,999][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:26:31,999][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:26:32,614][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:26:33,069][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:26:33,572][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:26:34,071][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:26:34,571][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:26:35,070][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:26:35,569][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:26:36,068][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:26:36,567][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:26:37,069][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:26:37,571][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:26:38,071][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:26:38,571][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:26:39,071][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:26:39,571][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:26:40,072][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:26:40,572][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:26:41,073][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:26:41,574][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:26:42,073][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:26:42,573][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:26:43,074][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:26:43,574][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:26:44,074][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:26:44,576][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:26:45,076][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:26:45,578][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:26:46,079][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:26:46,579][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:26:47,078][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:26:47,577][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:26:48,077][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:26:48,579][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:26:49,079][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:26:49,578][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:26:50,077][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:26:50,577][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:26:51,077][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:26:51,578][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:26:52,078][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:26:52,577][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:26:53,077][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:26:53,577][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:26:54,077][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:26:54,578][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:26:55,080][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:26:55,581][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:26:56,080][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:26:56,581][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:26:57,080][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:26:57,579][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:26:58,078][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:26:58,579][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:26:59,079][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:26:59,578][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:27:00,077][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:27:00,577][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:27:01,076][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:27:01,576][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:27:02,076][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:27:02,576][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:27:03,076][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:27:03,575][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:27:04,076][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:27:04,576][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:27:05,076][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:27:05,576][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:27:06,075][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:27:06,574][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:27:07,072][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:27:07,571][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:27:08,070][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:27:08,570][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:27:09,069][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:27:09,568][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:27:10,067][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:27:10,566][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:27:11,065][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:27:11,563][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:27:12,061][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:27:12,559][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:27:13,059][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:27:13,559][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:27:14,059][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:27:14,558][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:27:15,057][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:27:15,558][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:27:16,058][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:27:16,558][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:27:17,058][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:27:17,556][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:27:18,055][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:27:18,555][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:27:19,054][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:27:19,553][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:27:20,052][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:27:20,551][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:27:21,051][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:27:21,550][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:27:22,049][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:27:22,547][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:27:23,046][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:27:23,544][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:27:24,042][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:27:24,541][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:27:25,041][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:27:25,542][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:27:26,041][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:27:26,541][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:27:27,041][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:27:27,540][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:27:28,040][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:27:28,539][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:27:29,040][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:27:29,539][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:27:30,038][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:27:30,537][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:27:31,037][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:27:31,537][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:27:32,037][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:27:32,537][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:27:33,036][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:27:33,536][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:27:34,035][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:27:34,534][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:27:35,034][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:27:35,533][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:27:36,032][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:27:36,531][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21741 tokens. [2026-03-26 04:27:37,194][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:27:37,931][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:27:37,933][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:27:37,935][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:27:38,618][__main__][INFO] - Iteration 372 took 1m 14s (9.54% Gen, 89.55% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 54h 20m 2s. Estimated total time: 62h 12m 58s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 25s, 500 more iterations: 10h 22m 9s. [2026-03-26 04:27:38,621][__main__][INFO] - Starting iteration 372. [2026-03-26 04:27:39,023][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:27:39,023][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:27:39,609][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:27:46,066][__main__][INFO] - Number of regex retries in iteration 372: 1 [2026-03-26 04:27:46,067][__main__][INFO] - agents played in iteration 372 are Alice, Bob [2026-03-26 04:27:46,947][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:27:46,991][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:27:46,992][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:27:46,993][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:27:47,603][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:27:48,060][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:27:48,565][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:27:49,064][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:27:49,564][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:27:50,062][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:27:50,561][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:27:51,060][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:27:51,559][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:27:52,060][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:27:52,561][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:27:53,061][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:27:53,561][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:27:54,061][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:27:54,561][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:27:55,062][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:27:55,562][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:27:56,060][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:27:56,558][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:27:57,055][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:27:57,553][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:27:58,052][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:27:58,550][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:27:59,051][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:27:59,550][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:28:00,049][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:28:00,548][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:28:01,047][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:28:01,546][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:28:02,044][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:28:02,544][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:28:03,043][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:28:03,542][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:28:04,042][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:28:04,541][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:28:05,040][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:28:05,540][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:28:06,038][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:28:06,538][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:28:07,038][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:28:07,539][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:28:08,037][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:28:08,535][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:28:09,033][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:28:09,531][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:28:10,030][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:28:10,529][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:28:11,027][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:28:11,526][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:28:12,025][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:28:12,524][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:28:13,026][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:28:13,525][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:28:14,024][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:28:14,523][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:28:15,022][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:28:15,524][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:28:16,026][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:28:16,526][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:28:17,025][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:28:17,527][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:28:18,028][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:28:18,531][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:28:19,030][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:28:19,529][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:28:20,028][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:28:20,526][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:28:21,026][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:28:21,524][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:28:22,023][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:28:22,522][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:28:23,021][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:28:23,519][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:28:24,019][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:28:24,518][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:28:25,018][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:28:25,519][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:28:26,019][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:28:26,519][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:28:27,018][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:28:27,518][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:28:28,016][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:28:28,514][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:28:29,012][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:28:29,510][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:28:30,009][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:28:30,506][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:28:31,003][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:28:31,500][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:28:32,000][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:28:32,498][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:28:32,996][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:28:33,496][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:28:33,995][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:28:34,493][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:28:34,992][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:28:35,490][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:28:35,990][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:28:36,488][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:28:36,989][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:28:37,489][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:28:37,989][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:28:38,489][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:28:38,988][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:28:39,487][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:28:39,985][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:28:40,483][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:28:40,982][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:28:41,481][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:28:41,980][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:28:42,478][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:28:42,976][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:28:43,476][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:28:43,975][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:28:44,474][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:28:44,974][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:28:45,473][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:28:45,972][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:28:46,471][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:28:46,970][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:28:47,470][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:28:47,971][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:28:48,473][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:28:48,973][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:28:49,473][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:28:49,973][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:28:50,472][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:28:50,973][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:28:51,473][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:28:52,120][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 04:28:52,853][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:28:52,855][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:28:52,857][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:28:53,585][__main__][INFO] - Iteration 373 took 1m 14s (9.45% Gen, 89.57% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 54h 13m 58s. Estimated total time: 62h 8m 9s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 16s, 500 more iterations: 10h 21m 21s. [2026-03-26 04:28:53,588][__main__][INFO] - Starting iteration 373. [2026-03-26 04:28:53,996][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:28:53,997][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:29:00,293][__main__][INFO] - Number of regex retries in iteration 373: 0 [2026-03-26 04:29:00,294][__main__][INFO] - agents played in iteration 373 are Alice, Bob [2026-03-26 04:29:01,174][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:29:01,495][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:29:01,496][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:29:01,497][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:29:02,098][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:29:02,555][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:29:03,060][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:29:03,559][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:29:04,058][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:29:04,558][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:29:05,059][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:29:05,558][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:29:06,059][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:29:06,558][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:29:07,056][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:29:07,555][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:29:08,053][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:29:08,551][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:29:09,052][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:29:09,551][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:29:10,049][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:29:10,548][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:29:11,046][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:29:11,544][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:29:12,042][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:29:12,539][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:29:13,038][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:29:13,537][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:29:14,035][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:29:14,534][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:29:15,032][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:29:15,531][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:29:16,029][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:29:16,527][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:29:17,025][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:29:17,525][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:29:18,023][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:29:18,522][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:29:19,020][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:29:19,518][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:29:20,016][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:29:20,514][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:29:21,013][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:29:21,511][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:29:22,010][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:29:22,511][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:29:23,010][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:29:23,508][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:29:24,006][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:29:24,504][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:29:25,002][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:29:25,501][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:29:25,999][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:29:26,499][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:29:26,998][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:29:27,497][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:29:27,996][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:29:28,495][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:29:28,994][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:29:29,493][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:29:29,993][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:29:30,491][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:29:30,991][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:29:31,489][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:29:31,988][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:29:32,486][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:29:32,984][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:29:33,482][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:29:33,980][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:29:34,480][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:29:34,980][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:29:35,480][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:29:35,979][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:29:36,478][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:29:36,978][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:29:37,478][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:29:37,977][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:29:38,476][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:29:38,975][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:29:39,474][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:29:39,973][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:29:40,471][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:29:40,971][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:29:41,471][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:29:41,969][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:29:42,469][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:29:42,967][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:29:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:29:43,963][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:29:44,461][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:29:44,960][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:29:45,462][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:29:45,964][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:29:46,462][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:29:46,961][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:29:47,459][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:29:47,958][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:29:48,458][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:29:48,957][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:29:49,456][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:29:49,955][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:29:50,452][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:29:50,951][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:29:51,449][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:29:51,947][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:29:52,445][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:29:52,943][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:29:53,441][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:29:53,939][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:29:54,438][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:29:54,937][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:29:55,436][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:29:55,934][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:29:56,433][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:29:56,933][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:29:57,432][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:29:57,930][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:29:58,430][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:29:58,929][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:29:59,428][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:29:59,926][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:30:00,425][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:30:00,926][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:30:01,424][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:30:01,924][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:30:02,421][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:30:02,919][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:30:03,417][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:30:03,916][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:30:04,414][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:30:04,914][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:30:05,413][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:30:05,912][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 04:30:06,547][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 04:30:07,309][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:30:07,312][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:30:07,313][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:30:08,051][__main__][INFO] - Iteration 374 took 1m 14s (8.50% Gen, 90.50% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 47m 20s. Estimated total time: 61h 42m 46s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 25s, 500 more iterations: 10h 17m 7s. [2026-03-26 04:30:08,054][__main__][INFO] - Starting iteration 374. [2026-03-26 04:30:08,453][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:30:08,454][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:30:14,963][__main__][INFO] - Number of regex retries in iteration 374: 0 [2026-03-26 04:30:14,964][__main__][INFO] - agents played in iteration 374 are Alice, Bob [2026-03-26 04:30:15,841][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:30:15,885][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:30:15,886][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:30:15,886][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:30:16,485][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:30:16,940][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:30:17,443][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:30:17,942][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:30:18,443][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:30:18,941][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:30:19,440][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:30:19,940][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:30:20,439][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:30:20,940][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:30:21,440][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:30:21,940][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:30:22,439][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:30:22,938][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:30:23,437][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:30:23,937][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:30:24,437][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:30:24,937][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:30:25,438][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:30:25,937][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:30:26,437][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:30:26,937][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:30:27,436][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:30:27,935][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:30:28,435][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:30:28,934][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:30:29,432][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:30:29,929][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:30:30,426][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:30:30,924][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:30:31,422][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:30:31,921][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:30:32,419][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:30:32,917][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:30:33,415][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:30:33,913][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:30:34,411][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:30:34,908][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:30:35,405][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:30:35,902][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:30:36,400][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:30:36,900][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:30:37,399][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:30:37,897][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:30:38,396][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:30:38,895][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:30:39,393][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:30:39,891][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:30:40,388][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:30:40,888][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:30:41,386][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:30:41,885][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:30:42,383][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:30:42,882][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:30:43,380][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:30:43,877][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:30:44,376][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:30:44,874][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:30:45,373][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:30:45,871][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:30:46,370][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:30:46,870][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:30:47,368][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:30:47,866][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:30:48,366][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:30:48,865][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:30:49,363][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:30:49,860][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:30:50,359][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:30:50,858][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:30:51,356][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:30:51,855][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:30:52,354][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:30:52,852][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:30:53,353][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:30:53,851][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:30:54,351][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:30:54,851][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:30:55,350][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:30:55,848][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:30:56,346][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:30:56,844][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:30:57,344][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:30:57,843][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:30:58,342][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:30:58,839][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:30:59,337][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:30:59,837][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:31:00,335][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:31:00,833][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:31:01,332][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:31:01,831][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:31:02,329][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:31:02,826][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:31:03,323][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:31:03,821][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:31:04,319][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:31:04,817][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:31:05,315][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:31:05,814][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:31:06,312][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:31:06,811][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:31:07,310][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:31:07,808][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:31:08,306][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:31:08,805][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:31:09,304][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:31:09,803][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:31:10,302][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:31:10,801][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:31:11,300][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:31:11,799][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:31:12,298][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:31:12,797][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:31:13,295][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:31:13,795][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:31:14,293][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:31:14,791][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:31:15,292][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:31:15,793][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:31:16,292][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:31:16,790][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:31:17,288][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:31:17,786][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:31:18,287][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:31:18,785][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:31:19,283][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:31:19,783][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:31:20,281][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21736 tokens. [2026-03-26 04:31:20,903][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:31:21,650][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:31:21,652][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:31:21,654][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:31:22,313][__main__][INFO] - Iteration 375 took 1m 13s (8.81% Gen, 90.29% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 36m 20s. Estimated total time: 61h 33m 0s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 6s, 500 more iterations: 10h 15m 30s. [2026-03-26 04:31:22,315][__main__][INFO] - Starting iteration 375. [2026-03-26 04:31:22,718][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:31:22,719][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:31:23,787][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:31:27,970][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:31:29,551][__main__][INFO] - Number of regex retries in iteration 375: 2 [2026-03-26 04:31:29,552][__main__][INFO] - agents played in iteration 375 are Alice, Bob [2026-03-26 04:31:30,424][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:31:30,469][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:31:30,469][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:31:30,470][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:31:31,071][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:31:31,527][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:31:32,031][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:31:32,532][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:31:33,032][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:31:33,534][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:31:34,034][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:31:34,534][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:31:35,034][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:31:35,534][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:31:36,032][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:31:36,530][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:31:37,031][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:31:37,531][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:31:38,030][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:31:38,529][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:31:39,028][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:31:39,529][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:31:40,028][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:31:40,530][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:31:41,031][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:31:41,532][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:31:42,032][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:31:42,533][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:31:43,034][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:31:43,534][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:31:44,035][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:31:44,535][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:31:45,035][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:31:45,535][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:31:46,035][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:31:46,535][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:31:47,036][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:31:47,536][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:31:48,037][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:31:48,539][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:31:49,039][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:31:49,538][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:31:50,037][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:31:50,538][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:31:51,038][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:31:51,537][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:31:52,036][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:31:52,534][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:31:53,033][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:31:53,531][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:31:54,030][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:31:54,529][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:31:55,028][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:31:55,528][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:31:56,027][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:31:56,527][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:31:57,026][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:31:57,524][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:31:58,023][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:31:58,522][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:31:59,020][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:31:59,519][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:32:00,018][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:32:00,517][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:32:01,016][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:32:01,515][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:32:02,014][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:32:02,513][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:32:03,013][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:32:03,513][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:32:04,013][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:32:04,512][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:32:05,012][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:32:05,511][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:32:06,013][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:32:06,514][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:32:07,015][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:32:07,513][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:32:08,011][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:32:08,509][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:32:09,006][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:32:09,505][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:32:10,003][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:32:10,501][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:32:11,000][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:32:11,501][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:32:12,001][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:32:12,501][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:32:13,002][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:32:13,502][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:32:14,003][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:32:14,503][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:32:15,004][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:32:15,503][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:32:16,003][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:32:16,503][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:32:17,003][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:32:17,503][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:32:18,003][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:32:18,506][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:32:19,005][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:32:19,504][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:32:20,003][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:32:20,502][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:32:21,003][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:32:21,503][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:32:22,005][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:32:22,504][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:32:23,003][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:32:23,501][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:32:24,000][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:32:24,500][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:32:24,998][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:32:25,497][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:32:25,997][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:32:26,495][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:32:26,993][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:32:27,493][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:32:27,993][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:32:28,493][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:32:28,993][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:32:29,492][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:32:29,991][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:32:30,490][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:32:30,988][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:32:31,488][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:32:31,987][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:32:32,486][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:32:32,986][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:32:33,484][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:32:33,983][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:32:34,482][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:32:34,981][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 04:32:35,617][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 04:32:37,287][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:32:37,290][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:32:37,291][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:32:38,154][__main__][INFO] - Iteration 376 took 1m 15s (9.06% Gen, 89.80% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 54h 53m 54s. Estimated total time: 62h 51m 50s. Time estimates for 10 more iterations: 12m 34s, 100 more iterations: 2h 5m 43s, 500 more iterations: 10h 28m 38s. [2026-03-26 04:32:38,157][__main__][INFO] - Starting iteration 376. [2026-03-26 04:32:38,555][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:32:38,556][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:32:45,215][__main__][INFO] - Number of regex retries in iteration 376: 0 [2026-03-26 04:32:45,216][__main__][INFO] - agents played in iteration 376 are Alice, Bob [2026-03-26 04:32:46,091][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:32:46,136][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:32:46,136][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:32:46,137][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:32:46,751][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:32:47,211][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:32:47,715][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:32:48,215][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:32:48,716][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:32:49,218][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:32:49,717][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:32:50,217][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:32:50,717][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:32:51,216][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:32:51,715][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:32:52,216][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:32:52,717][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:32:53,217][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:32:53,717][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:32:54,216][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:32:54,717][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:32:55,218][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:32:55,717][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:32:56,217][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:32:56,717][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:32:57,216][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:32:57,716][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:32:58,217][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:32:58,717][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:32:59,216][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:32:59,715][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:33:00,214][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:33:00,712][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:33:01,210][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:33:01,708][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:33:02,206][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:33:02,704][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:33:03,204][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:33:03,703][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:33:04,203][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:33:04,702][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:33:05,202][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:33:05,701][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:33:06,201][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:33:06,701][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:33:07,200][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:33:07,699][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:33:08,198][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:33:08,697][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:33:09,196][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:33:09,696][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:33:10,195][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:33:10,694][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:33:11,194][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:33:11,693][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:33:12,193][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:33:12,691][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:33:13,190][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:33:13,688][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:33:14,187][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:33:14,686][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:33:15,184][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:33:15,683][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:33:16,181][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:33:16,678][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:33:17,176][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:33:17,676][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:33:18,177][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:33:18,677][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:33:19,177][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:33:19,677][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:33:20,176][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:33:20,675][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:33:21,175][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:33:21,676][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:33:22,176][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:33:22,678][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:33:23,178][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:33:23,677][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:33:24,176][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:33:24,675][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:33:25,174][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:33:25,674][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:33:26,173][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:33:26,672][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:33:27,171][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:33:27,669][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:33:28,167][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:33:28,665][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:33:29,162][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:33:29,661][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:33:30,160][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:33:30,658][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:33:31,157][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:33:31,654][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:33:32,153][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:33:32,650][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:33:33,148][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:33:33,645][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:33:34,142][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:33:34,640][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:33:35,139][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:33:35,638][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:33:36,137][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:33:36,636][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:33:37,134][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:33:37,634][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:33:38,133][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:33:38,633][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:33:39,132][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:33:39,632][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:33:40,130][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:33:40,629][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:33:41,126][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:33:41,625][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:33:42,124][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:33:42,622][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:33:43,121][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:33:43,621][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:33:44,119][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:33:44,618][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:33:45,117][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:33:45,616][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:33:46,115][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:33:46,614][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:33:47,112][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:33:47,611][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:33:48,109][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:33:48,610][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:33:49,107][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:33:49,606][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:33:50,103][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:33:50,601][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 04:33:51,234][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 04:33:52,034][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:33:52,036][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:33:52,038][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:33:52,808][__main__][INFO] - Iteration 377 took 1m 14s (8.97% Gen, 89.99% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 53m 29s. Estimated total time: 61h 52m 40s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 45s, 500 more iterations: 10h 18m 46s. [2026-03-26 04:33:52,810][__main__][INFO] - Starting iteration 377. [2026-03-26 04:33:53,211][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:33:53,212][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:33:55,949][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:33:59,796][__main__][INFO] - Number of regex retries in iteration 377: 1 [2026-03-26 04:33:59,797][__main__][INFO] - agents played in iteration 377 are Alice, Bob [2026-03-26 04:34:00,674][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:34:00,718][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:34:00,719][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:34:00,719][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:34:01,333][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:34:01,789][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:34:02,294][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:34:02,794][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:34:03,294][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:34:03,795][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:34:04,294][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:34:04,793][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:34:05,294][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:34:05,795][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:34:06,295][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:34:06,795][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:34:07,293][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:34:07,793][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:34:08,291][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:34:08,793][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:34:09,293][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:34:09,792][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:34:10,291][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:34:10,788][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:34:11,286][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:34:11,784][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:34:12,282][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:34:12,780][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:34:13,279][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:34:13,781][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:34:14,282][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:34:14,781][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:34:15,282][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:34:15,786][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:34:16,286][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:34:16,785][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:34:17,285][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:34:17,785][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:34:18,286][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:34:18,786][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:34:19,285][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:34:19,784][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:34:20,283][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:34:20,782][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:34:21,282][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:34:21,782][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:34:22,282][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:34:22,783][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:34:23,282][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:34:23,784][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:34:24,285][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:34:24,784][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:34:25,285][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:34:25,784][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:34:26,283][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:34:26,783][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:34:27,283][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:34:27,784][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:34:28,284][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:34:28,784][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:34:29,284][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:34:29,780][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:34:30,277][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:34:30,775][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:34:31,273][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:34:31,770][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:34:32,267][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:34:32,764][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:34:33,260][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:34:33,760][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:34:34,258][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:34:34,758][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:34:35,257][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:34:35,757][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:34:36,255][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:34:36,754][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:34:37,253][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:34:37,752][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:34:38,252][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:34:38,752][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:34:39,252][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:34:39,752][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:34:40,252][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:34:40,750][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:34:41,250][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:34:41,748][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:34:42,247][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:34:42,744][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:34:43,242][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:34:43,741][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:34:44,239][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:34:44,737][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:34:45,236][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:34:45,736][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:34:46,236][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:34:46,736][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:34:47,235][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:34:47,735][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:34:48,237][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:34:48,739][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:34:49,238][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:34:49,738][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:34:50,237][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:34:50,736][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:34:51,236][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:34:51,735][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:34:52,235][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:34:52,734][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:34:53,234][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:34:53,735][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:34:54,235][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:34:54,734][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:34:55,234][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:34:55,734][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:34:56,233][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:34:56,732][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:34:57,234][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:34:57,734][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:34:58,234][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:34:58,733][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:34:59,233][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:34:59,733][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:35:00,233][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:35:00,733][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:35:01,233][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:35:01,730][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:35:02,226][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:35:02,723][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:35:03,220][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:35:03,717][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:35:04,214][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:35:04,709][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:35:05,206][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:35:05,846][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.14%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:35:06,559][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:35:06,562][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:35:06,563][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:35:07,208][__main__][INFO] - Iteration 378 took 1m 13s (8.90% Gen, 90.23% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 39m 24s. Estimated total time: 61h 39m 49s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 19s, 500 more iterations: 10h 16m 38s. [2026-03-26 04:35:07,210][__main__][INFO] - Starting iteration 378. [2026-03-26 04:35:07,613][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:35:07,614][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:35:09,348][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 balls, 10 books did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:35:11,279][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:35:14,751][__main__][INFO] - Number of regex retries in iteration 378: 2 [2026-03-26 04:35:14,752][__main__][INFO] - agents played in iteration 378 are Alice, Bob [2026-03-26 04:35:15,652][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:35:15,696][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:35:15,697][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:35:15,698][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:35:16,308][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:35:16,760][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:35:17,262][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:35:17,761][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:35:18,262][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:35:18,760][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:35:19,258][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:35:19,755][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:35:20,253][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:35:20,756][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:35:21,255][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:35:21,753][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:35:22,251][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:35:22,749][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:35:23,247][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:35:23,746][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:35:24,244][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:35:24,745][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:35:25,245][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:35:25,744][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:35:26,243][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:35:26,742][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:35:27,240][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:35:27,739][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:35:28,238][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:35:28,740][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:35:29,241][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:35:29,740][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:35:30,240][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:35:30,740][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:35:31,240][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:35:31,740][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:35:32,240][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:35:32,739][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:35:33,237][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:35:33,735][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:35:34,234][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:35:34,733][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:35:35,231][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:35:35,729][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:35:36,227][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:35:36,724][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:35:37,221][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:35:37,718][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:35:38,215][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:35:38,712][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:35:39,209][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:35:39,705][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:35:40,201][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:35:40,702][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:35:41,201][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:35:41,702][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:35:42,202][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:35:42,702][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:35:43,202][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:35:43,702][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:35:44,202][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:35:44,703][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:35:45,202][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:35:45,705][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:35:46,206][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:35:46,705][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:35:47,205][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:35:47,707][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:35:48,210][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:35:48,710][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:35:49,208][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:35:49,706][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:35:50,205][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:35:50,702][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:35:51,200][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:35:51,698][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:35:52,195][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:35:52,694][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:35:53,193][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:35:53,692][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:35:54,190][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:35:54,688][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:35:55,187][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:35:55,685][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:35:56,183][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:35:56,683][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:35:57,182][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:35:57,680][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:35:58,179][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:35:58,678][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:35:59,176][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:35:59,676][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:36:00,175][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:36:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:36:01,176][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:36:01,676][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:36:02,177][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:36:02,678][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:36:03,177][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:36:03,677][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:36:04,178][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:36:04,677][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:36:05,176][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:36:05,675][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:36:06,174][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:36:06,674][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:36:07,173][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:36:07,671][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:36:08,170][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:36:08,667][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:36:09,164][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:36:09,661][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:36:10,159][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:36:10,656][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:36:11,155][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:36:11,652][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:36:12,149][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:36:12,650][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:36:13,149][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:36:13,649][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:36:14,148][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:36:14,649][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:36:15,151][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:36:15,651][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:36:16,151][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:36:16,651][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:36:17,150][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:36:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:36:18,151][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:36:18,653][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:36:19,153][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:36:19,653][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:36:20,152][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:36:20,789][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 04:36:21,627][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:36:21,629][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:36:21,630][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:36:22,339][__main__][INFO] - Iteration 379 took 1m 14s (9.55% Gen, 89.50% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 54h 14m 39s. Estimated total time: 62h 16m 19s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 32s, 500 more iterations: 10h 22m 43s. [2026-03-26 04:36:22,341][__main__][INFO] - Starting iteration 379. [2026-03-26 04:36:22,747][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:36:22,748][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:36:23,334][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:36:25,527][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:36:26,487][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:36:28,983][__main__][INFO] - Number of regex retries in iteration 379: 3 [2026-03-26 04:36:28,984][__main__][INFO] - agents played in iteration 379 are Alice, Bob [2026-03-26 04:36:29,861][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:36:29,906][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:36:29,907][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:36:29,907][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:36:30,518][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:36:30,973][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:36:31,479][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:36:31,980][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:36:32,481][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:36:32,982][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:36:33,482][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:36:33,982][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:36:34,481][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:36:34,980][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:36:35,479][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:36:35,979][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:36:36,478][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:36:36,977][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:36:37,476][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:36:37,975][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:36:38,474][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:36:38,972][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:36:39,470][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:36:39,968][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:36:40,466][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:36:40,963][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:36:41,461][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:36:41,960][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:36:42,459][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:36:42,959][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:36:43,458][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:36:43,957][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:36:44,458][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:36:44,959][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:36:45,460][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:36:45,961][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:36:46,460][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:36:46,959][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:36:47,458][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:36:47,959][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:36:48,462][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:36:48,961][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:36:49,460][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:36:49,959][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:36:50,458][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:36:50,957][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:36:51,456][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:36:51,954][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:36:52,455][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:36:52,954][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:36:53,454][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:36:53,953][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:36:54,452][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:36:54,955][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:36:55,458][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:36:55,960][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:36:56,463][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:36:56,964][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:36:57,466][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:36:57,968][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:36:58,471][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:36:58,972][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:36:59,471][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:36:59,970][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:37:00,468][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:37:00,967][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:37:01,465][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:37:01,962][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:37:02,461][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:37:02,961][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:37:03,460][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:37:03,959][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:37:04,458][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:37:04,958][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:37:05,458][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:37:05,956][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:37:06,458][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:37:06,957][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:37:07,457][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:37:07,955][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:37:08,454][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:37:08,953][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:37:09,452][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:37:09,953][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:37:10,453][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:37:10,952][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:37:11,450][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:37:11,947][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:37:12,445][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:37:12,943][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:37:13,440][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:37:13,937][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:37:14,436][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:37:14,935][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:37:15,437][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:37:15,936][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:37:16,436][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:37:16,938][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:37:17,437][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:37:17,938][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:37:18,439][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:37:18,938][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:37:19,436][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:37:19,936][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:37:20,434][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:37:20,931][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:37:21,430][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:37:21,928][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:37:22,427][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:37:22,925][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:37:23,424][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:37:23,922][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:37:24,421][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:37:24,919][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:37:25,420][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:37:25,920][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:37:26,420][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:37:26,921][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:37:27,423][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:37:27,925][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:37:28,427][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:37:28,928][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:37:29,430][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:37:29,932][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:37:30,433][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:37:30,931][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:37:31,429][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:37:31,927][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:37:32,426][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:37:32,924][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:37:33,422][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:37:33,921][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:37:34,419][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 04:37:35,060][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 04:37:35,834][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:37:35,837][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:37:35,839][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:37:36,535][__main__][INFO] - Iteration 380 took 1m 13s (8.45% Gen, 90.60% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 26m 32s. Estimated total time: 61h 29m 26s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 58s, 500 more iterations: 10h 14m 54s. [2026-03-26 04:37:36,537][__main__][INFO] - Starting iteration 380. [2026-03-26 04:37:36,940][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:37:36,940][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:37:43,848][__main__][INFO] - Number of regex retries in iteration 380: 0 [2026-03-26 04:37:43,849][__main__][INFO] - agents played in iteration 380 are Alice, Bob [2026-03-26 04:37:44,730][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:37:44,774][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:37:44,775][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:37:44,775][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:37:45,380][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:37:45,834][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:37:46,336][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:37:46,835][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:37:47,334][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:37:47,832][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:37:48,334][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:37:48,832][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:37:49,329][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:37:49,831][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:37:50,330][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:37:50,829][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:37:51,329][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:37:51,828][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:37:52,327][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:37:52,826][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:37:53,325][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:37:53,823][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:37:54,322][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:37:54,820][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:37:55,320][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:37:55,819][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:37:56,317][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:37:56,816][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:37:57,315][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:37:57,813][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:37:58,310][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:37:58,807][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:37:59,305][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:37:59,803][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:38:00,301][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:38:00,799][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:38:01,297][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:38:01,796][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:38:02,295][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:38:02,793][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:38:03,294][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:38:03,794][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:38:04,293][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:38:04,793][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:38:05,291][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:38:05,809][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:38:06,315][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:38:06,815][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:38:07,316][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:38:07,816][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:38:08,317][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:38:08,817][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:38:09,318][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:38:09,817][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:38:10,317][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:38:10,815][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:38:11,314][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:38:11,813][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:38:12,311][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:38:12,810][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:38:13,310][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:38:13,810][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:38:14,309][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:38:14,809][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:38:15,309][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:38:15,808][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:38:16,308][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:38:16,807][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:38:17,307][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:38:17,806][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:38:18,305][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:38:18,804][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:38:19,301][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:38:19,799][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:38:20,296][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:38:20,795][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:38:21,293][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:38:21,792][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:38:22,292][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:38:22,791][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:38:23,290][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:38:23,788][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:38:24,287][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:38:24,787][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:38:25,287][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:38:25,786][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:38:26,286][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:38:26,784][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:38:27,282][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:38:27,780][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:38:28,278][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:38:28,778][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:38:29,277][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:38:29,775][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:38:30,274][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:38:30,771][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:38:31,269][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:38:31,766][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:38:32,264][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:38:32,762][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:38:33,261][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:38:33,761][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:38:34,261][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:38:34,760][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:38:35,261][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:38:35,761][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:38:36,261][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:38:36,761][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:38:37,262][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:38:37,762][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:38:38,262][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:38:38,763][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:38:39,263][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:38:39,764][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:38:40,265][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:38:40,765][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:38:41,265][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:38:41,765][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:38:42,263][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:38:42,762][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:38:43,262][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:38:43,761][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:38:44,261][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:38:44,760][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:38:45,260][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:38:45,763][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:38:46,264][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:38:46,764][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:38:47,264][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:38:47,763][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:38:48,265][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:38:48,766][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:38:49,265][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 04:38:49,893][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 04:38:50,781][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:38:50,784][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:38:50,785][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:38:51,492][__main__][INFO] - Iteration 381 took 1m 14s (9.27% Gen, 89.78% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 3m 30s. Estimated total time: 62h 7m 39s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 15s, 500 more iterations: 10h 21m 16s. [2026-03-26 04:38:51,494][__main__][INFO] - Starting iteration 381. [2026-03-26 04:38:51,893][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:38:51,894][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:38:53,557][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:38:58,118][__main__][INFO] - Number of regex retries in iteration 381: 1 [2026-03-26 04:38:58,119][__main__][INFO] - agents played in iteration 381 are Alice, Bob [2026-03-26 04:38:59,004][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:38:59,049][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:38:59,050][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:38:59,050][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:38:59,655][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:39:00,113][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:39:00,616][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:39:01,116][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:39:01,616][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:39:02,116][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:39:02,615][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:39:03,114][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:39:03,614][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:39:04,112][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:39:04,610][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:39:05,108][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:39:05,606][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:39:06,105][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:39:06,604][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:39:07,103][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:39:07,602][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:39:08,103][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:39:08,603][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:39:09,102][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:39:09,602][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:39:10,101][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:39:10,601][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:39:11,100][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:39:11,601][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:39:12,102][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:39:12,601][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:39:13,101][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:39:13,601][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:39:14,101][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:39:14,600][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:39:15,101][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:39:15,599][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:39:16,098][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:39:16,598][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:39:17,098][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:39:17,598][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:39:18,098][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:39:18,601][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:39:19,101][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:39:19,601][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:39:20,100][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:39:20,600][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:39:21,099][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:39:21,598][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:39:22,097][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:39:22,596][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:39:23,097][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:39:23,597][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:39:24,096][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:39:24,595][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:39:25,094][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:39:25,593][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:39:26,091][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:39:26,589][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:39:27,089][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:39:27,589][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:39:28,085][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:39:28,583][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:39:29,080][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:39:29,578][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:39:30,076][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:39:30,573][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:39:31,070][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:39:31,568][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:39:32,067][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:39:32,567][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:39:33,066][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:39:33,566][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:39:34,065][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:39:34,565][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:39:35,064][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:39:35,564][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:39:36,063][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:39:36,561][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:39:37,060][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:39:37,558][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:39:38,057][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:39:38,555][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:39:39,056][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:39:39,555][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:39:40,054][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:39:40,554][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:39:41,053][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:39:41,552][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:39:42,051][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:39:42,551][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:39:43,050][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:39:43,550][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:39:44,050][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:39:44,549][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:39:45,049][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:39:45,548][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:39:46,046][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:39:46,546][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:39:47,046][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:39:47,545][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:39:48,044][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:39:48,544][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:39:49,044][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:39:49,544][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:39:50,043][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:39:50,543][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:39:51,043][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:39:51,543][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:39:52,042][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:39:52,541][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:39:53,040][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:39:53,539][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:39:54,038][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:39:54,537][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:39:55,037][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:39:55,536][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:39:56,034][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:39:56,532][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:39:57,030][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:39:57,528][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:39:58,026][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:39:58,525][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:39:59,024][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:39:59,523][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:40:00,019][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:40:00,515][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:40:01,012][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:40:01,508][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:40:02,004][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:40:02,501][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:40:02,999][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:40:03,497][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21739 tokens. [2026-03-26 04:40:04,123][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:40:05,158][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:40:05,161][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:40:05,162][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:40:05,842][__main__][INFO] - Iteration 382 took 1m 13s (8.42% Gen, 90.66% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 32m 5s. Estimated total time: 61h 37m 28s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 14s, 500 more iterations: 10h 16m 14s. [2026-03-26 04:40:05,844][__main__][INFO] - Starting iteration 382. [2026-03-26 04:40:06,244][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:40:06,245][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:40:12,943][__main__][INFO] - Number of regex retries in iteration 382: 0 [2026-03-26 04:40:12,944][__main__][INFO] - agents played in iteration 382 are Alice, Bob [2026-03-26 04:40:13,837][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:40:13,882][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:40:13,882][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:40:13,883][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:40:14,487][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:40:14,943][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:40:15,448][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:40:15,952][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:40:16,454][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:40:16,954][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:40:17,453][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:40:17,954][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:40:18,457][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:40:18,957][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:40:19,456][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:40:19,955][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:40:20,453][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:40:20,952][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:40:21,450][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:40:21,948][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:40:22,446][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:40:22,946][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:40:23,445][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:40:23,944][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:40:24,443][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:40:24,942][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:40:25,442][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:40:25,941][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:40:26,440][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:40:26,940][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:40:27,437][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:40:27,936][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:40:28,434][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:40:28,932][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:40:29,429][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:40:29,926][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:40:30,423][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:40:30,922][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:40:31,421][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:40:31,921][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:40:32,419][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:40:32,918][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:40:33,416][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:40:33,916][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:40:34,414][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:40:34,915][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:40:35,414][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:40:35,914][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:40:36,413][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:40:36,912][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:40:37,410][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:40:37,908][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:40:38,406][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:40:38,905][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:40:39,404][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:40:39,903][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:40:40,402][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:40:40,901][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:40:41,401][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:40:41,901][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:40:42,401][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:40:42,901][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:40:43,401][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:40:43,902][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:40:44,401][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:40:44,901][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:40:45,403][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:40:45,903][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:40:46,402][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:40:46,902][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:40:47,402][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:40:47,904][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:40:48,406][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:40:48,906][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:40:49,406][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:40:49,906][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:40:50,407][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:40:50,906][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:40:51,404][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:40:51,902][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:40:52,402][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:40:52,901][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:40:53,400][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:40:53,900][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:40:54,399][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:40:54,898][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:40:55,398][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:40:55,898][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:40:56,397][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:40:56,898][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:40:57,397][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:40:57,897][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:40:58,397][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:40:58,894][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:40:59,392][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:40:59,890][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:41:00,387][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:41:00,885][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:41:01,384][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:41:01,882][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:41:02,380][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:41:02,879][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:41:03,378][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:41:03,879][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:41:04,378][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:41:04,878][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:41:05,378][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:41:05,877][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:41:06,377][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:41:06,877][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:41:07,376][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:41:07,875][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:41:08,375][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:41:08,874][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:41:09,373][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:41:09,873][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:41:10,371][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:41:10,869][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:41:11,367][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:41:11,866][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:41:12,366][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:41:12,865][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:41:13,365][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:41:13,864][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:41:14,363][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:41:14,862][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:41:15,362][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:41:15,862][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:41:16,363][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:41:16,863][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:41:17,363][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:41:17,863][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:41:18,365][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 04:41:18,991][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:41:19,745][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:41:19,748][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:41:19,750][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:41:20,415][__main__][INFO] - Iteration 383 took 1m 14s (9.03% Gen, 90.07% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 41m 55s. Estimated total time: 61h 48m 34s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 37s, 500 more iterations: 10h 18m 5s. [2026-03-26 04:41:20,417][__main__][INFO] - Starting iteration 383. [2026-03-26 04:41:20,814][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:41:20,815][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:41:27,286][__main__][INFO] - Number of regex retries in iteration 383: 0 [2026-03-26 04:41:27,287][__main__][INFO] - agents played in iteration 383 are Alice, Bob [2026-03-26 04:41:28,161][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:41:28,482][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:41:28,483][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:41:28,483][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:41:29,088][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:41:29,543][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:41:30,047][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:41:30,546][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:41:31,045][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:41:31,545][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:41:32,045][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:41:32,545][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:41:33,044][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:41:33,545][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:41:34,045][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:41:34,544][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:41:35,044][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:41:35,544][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:41:36,044][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:41:36,545][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:41:37,044][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:41:37,543][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:41:38,042][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:41:38,542][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:41:39,041][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:41:39,539][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:41:40,038][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:41:40,536][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:41:41,034][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:41:41,533][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:41:42,032][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:41:42,531][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:41:43,030][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:41:43,529][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:41:44,029][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:41:44,527][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:41:45,027][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:41:45,528][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:41:46,028][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:41:46,527][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:41:47,026][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:41:47,525][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:41:48,025][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:41:48,526][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:41:49,024][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:41:49,523][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:41:50,024][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:41:50,523][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:41:51,023][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:41:51,523][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:41:52,022][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:41:52,521][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:41:53,021][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:41:53,521][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:41:54,020][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:41:54,519][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:41:55,019][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:41:55,519][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:41:56,017][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:41:56,517][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:41:57,016][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:41:57,517][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:41:58,016][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:41:58,515][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:41:59,015][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:41:59,513][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:42:00,010][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:42:00,508][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:42:01,007][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:42:01,506][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:42:02,005][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:42:02,505][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:42:03,004][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:42:03,504][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:42:04,002][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:42:04,502][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:42:05,002][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:42:05,503][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:42:06,003][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:42:06,503][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:42:07,004][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:42:07,504][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:42:08,004][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:42:08,505][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:42:09,005][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:42:09,504][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:42:10,003][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:42:10,502][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:42:11,001][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:42:11,500][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:42:11,998][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:42:12,497][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:42:12,996][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:42:13,497][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:42:13,997][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:42:14,497][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:42:14,997][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:42:15,496][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:42:15,998][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:42:16,497][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:42:16,997][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:42:17,496][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:42:17,996][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:42:18,497][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:42:18,996][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:42:19,497][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:42:19,997][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:42:20,496][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:42:20,996][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:42:21,495][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:42:21,994][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:42:22,493][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:42:22,994][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:42:23,496][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:42:23,996][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:42:24,495][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:42:24,996][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:42:25,495][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:42:25,993][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:42:26,493][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:42:26,993][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:42:27,493][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:42:27,993][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:42:28,492][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:42:28,991][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:42:29,489][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:42:29,990][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:42:30,488][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:42:30,989][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:42:31,489][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:42:31,987][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:42:32,486][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:42:32,985][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:42:33,615][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:42:34,351][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:42:34,354][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:42:34,355][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:42:35,055][__main__][INFO] - Iteration 384 took 1m 14s (8.72% Gen, 90.34% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 44m 9s. Estimated total time: 61h 52m 2s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 44s, 500 more iterations: 10h 18m 40s. [2026-03-26 04:42:35,057][__main__][INFO] - Starting iteration 384. [2026-03-26 04:42:35,455][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:42:35,456][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:42:36,045][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:42:42,191][__main__][INFO] - Number of regex retries in iteration 384: 1 [2026-03-26 04:42:42,192][__main__][INFO] - agents played in iteration 384 are Alice, Bob [2026-03-26 04:42:43,089][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:42:43,134][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:42:43,135][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:42:43,135][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:42:43,739][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:42:44,194][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:42:44,697][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:42:45,198][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:42:45,697][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:42:46,197][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:42:46,695][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:42:47,195][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:42:47,693][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:42:48,194][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:42:48,694][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:42:49,194][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:42:49,693][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:42:50,192][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:42:50,690][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:42:51,189][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:42:51,687][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:42:52,186][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:42:52,685][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:42:53,185][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:42:53,684][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:42:54,183][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:42:54,682][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:42:55,183][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:42:55,683][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:42:56,184][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:42:56,685][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:42:57,185][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:42:57,685][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:42:58,184][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:42:58,683][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:42:59,183][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:42:59,683][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:43:00,182][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:43:00,681][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:43:01,180][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:43:01,679][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:43:02,178][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:43:02,677][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:43:03,177][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:43:03,676][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:43:04,176][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:43:04,675][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:43:05,174][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:43:05,671][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:43:06,170][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:43:06,667][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:43:07,166][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:43:07,664][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:43:08,163][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:43:08,662][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:43:09,160][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:43:09,659][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:43:10,159][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:43:10,657][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:43:11,158][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:43:11,657][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:43:12,160][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:43:12,661][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:43:13,163][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:43:13,664][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:43:14,165][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:43:14,665][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:43:15,170][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:43:15,675][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:43:16,175][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:43:16,675][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:43:17,175][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:43:17,674][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:43:18,175][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:43:18,674][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:43:19,172][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:43:19,670][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:43:20,168][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:43:20,666][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:43:21,165][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:43:21,664][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:43:22,162][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:43:22,661][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:43:23,159][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:43:23,657][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:43:24,156][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:43:24,654][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:43:25,154][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:43:25,654][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:43:26,152][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:43:26,650][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:43:27,148][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:43:27,646][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:43:28,145][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:43:28,643][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:43:29,142][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:43:29,641][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:43:30,140][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:43:30,640][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:43:31,139][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:43:31,640][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:43:32,140][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:43:32,638][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:43:33,138][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:43:33,638][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:43:34,136][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:43:34,636][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:43:35,135][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:43:35,634][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:43:36,132][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:43:36,630][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:43:37,129][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:43:37,628][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:43:38,127][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:43:38,625][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:43:39,123][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:43:39,621][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:43:40,120][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:43:40,619][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:43:41,119][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:43:41,619][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:43:42,118][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:43:42,617][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:43:43,115][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:43:43,613][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:43:44,113][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:43:44,614][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:43:45,117][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:43:45,618][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:43:46,119][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:43:46,619][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:43:47,122][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:43:47,623][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21743 tokens. [2026-03-26 04:43:48,262][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.36%, Current % of VRAM taken: 60.71%, Block Peak % of device VRAM: 62.60%, ΔTime: 00:01:04 [2026-03-26 04:43:48,995][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:43:48,997][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:43:48,999][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:43:49,645][__main__][INFO] - Iteration 385 took 1m 14s (9.08% Gen, 90.05% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 40m 23s. Estimated total time: 61h 49m 30s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 15s. [2026-03-26 04:43:49,647][__main__][INFO] - Starting iteration 385. [2026-03-26 04:43:50,048][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:43:50,049][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:43:52,264][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:43:52,323][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:43:56,864][__main__][INFO] - Number of regex retries in iteration 385: 2 [2026-03-26 04:43:56,865][__main__][INFO] - agents played in iteration 385 are Alice, Bob [2026-03-26 04:43:57,737][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.54%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:43:57,781][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.54%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:43:57,782][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:43:57,783][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:43:58,389][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:43:58,844][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:43:59,347][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:43:59,847][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:44:00,345][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:44:00,844][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:44:01,343][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:44:01,843][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:44:02,343][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:44:02,841][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:44:03,339][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:44:03,837][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:44:04,335][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:44:04,831][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:44:05,328][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:44:05,826][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:44:06,324][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:44:06,824][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:44:07,324][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:44:07,823][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:44:08,322][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:44:08,822][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:44:09,321][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:44:09,821][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:44:10,320][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:44:10,821][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:44:11,321][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:44:11,821][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:44:12,320][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:44:12,819][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:44:13,319][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:44:13,818][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:44:14,317][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:44:14,817][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:44:15,316][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:44:15,815][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:44:16,313][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:44:16,811][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:44:17,310][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:44:17,809][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:44:18,311][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:44:18,811][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:44:19,311][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:44:19,810][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:44:20,310][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:44:20,809][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:44:21,308][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:44:21,809][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:44:22,309][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:44:22,808][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:44:23,306][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:44:23,805][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:44:24,305][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:44:24,804][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:44:25,304][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:44:25,804][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:44:26,303][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:44:26,803][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:44:27,303][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:44:27,804][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:44:28,304][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:44:28,804][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:44:29,304][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:44:29,804][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:44:30,303][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:44:30,803][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:44:31,302][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:44:31,804][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:44:32,303][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:44:32,805][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:44:33,304][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:44:33,803][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:44:34,302][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:44:34,800][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:44:35,299][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:44:35,798][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:44:36,297][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:44:36,796][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:44:37,294][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:44:37,792][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:44:38,290][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:44:38,788][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:44:39,288][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:44:39,787][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:44:40,286][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:44:40,785][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:44:41,284][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:44:41,783][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:44:42,283][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:44:42,784][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:44:43,284][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:44:43,783][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:44:44,283][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:44:44,783][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:44:45,283][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:44:45,787][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:44:46,288][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:44:46,787][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:44:47,287][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:44:47,786][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:44:48,288][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:44:48,788][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:44:49,287][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:44:49,787][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:44:50,286][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:44:50,786][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:44:51,285][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:44:51,785][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:44:52,285][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:44:52,784][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:44:53,283][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:44:53,784][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:44:54,284][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:44:54,783][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:44:55,283][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:44:55,782][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:44:56,281][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:44:56,781][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:44:57,280][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:44:57,780][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:44:58,279][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:44:58,778][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:44:59,279][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:44:59,779][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:45:00,278][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:45:00,778][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:45:01,278][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:45:01,778][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:45:02,279][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 04:45:02,910][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:45:03,832][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:45:03,835][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:45:03,836][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:45:04,689][__main__][INFO] - Iteration 386 took 1m 14s (9.13% Gen, 89.72% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 54h 1m 43s. Estimated total time: 62h 12m 6s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 1s. [2026-03-26 04:45:04,692][__main__][INFO] - Starting iteration 386. [2026-03-26 04:45:05,090][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:45:05,091][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:45:12,024][__main__][INFO] - Number of regex retries in iteration 386: 0 [2026-03-26 04:45:12,025][__main__][INFO] - agents played in iteration 386 are Alice, Bob [2026-03-26 04:45:12,901][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:45:12,945][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:45:12,946][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:45:12,946][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:45:13,555][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:45:14,011][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:45:14,518][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:45:15,018][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:45:15,516][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:45:16,014][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:45:16,512][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:45:17,011][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:45:17,509][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:45:18,009][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:45:18,510][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:45:19,009][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:45:19,508][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:45:20,007][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:45:20,505][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:45:21,004][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:45:21,503][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:45:22,002][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:45:22,501][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:45:23,000][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:45:23,499][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:45:23,997][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:45:24,496][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:45:24,997][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:45:25,496][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:45:25,996][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:45:26,495][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:45:26,993][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:45:27,492][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:45:27,990][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:45:28,488][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:45:28,986][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:45:29,485][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:45:29,984][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:45:30,482][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:45:30,980][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:45:31,479][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:45:31,978][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:45:32,477][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:45:32,979][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:45:33,480][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:45:33,979][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:45:34,479][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:45:34,977][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:45:35,476][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:45:35,974][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:45:36,473][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:45:36,972][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:45:37,469][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:45:37,969][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:45:38,468][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:45:38,966][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:45:39,464][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:45:39,963][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:45:40,462][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:45:40,962][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:45:41,462][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:45:41,962][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:45:42,462][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:45:42,963][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:45:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:45:43,965][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:45:44,468][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:45:44,968][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:45:45,468][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:45:45,967][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:45:46,467][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:45:46,968][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:45:47,469][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:45:47,970][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:45:48,472][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:45:48,972][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:45:49,470][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:45:49,965][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:45:50,464][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:45:50,964][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:45:51,464][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:45:51,963][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:45:52,462][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:45:52,960][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:45:53,460][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:45:53,960][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:45:54,458][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:45:54,959][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:45:55,459][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:45:55,958][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:45:56,458][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:45:56,958][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:45:57,457][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:45:57,955][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:45:58,455][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:45:58,954][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:45:59,453][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:45:59,951][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:46:00,449][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:46:00,947][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:46:01,445][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:46:01,943][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:46:02,441][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:46:02,940][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:46:03,439][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:46:03,937][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:46:04,437][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:46:04,935][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:46:05,433][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:46:05,930][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:46:06,429][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:46:06,928][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:46:07,426][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:46:07,925][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:46:08,424][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:46:08,924][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:46:09,423][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:46:09,922][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:46:10,421][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:46:10,920][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:46:11,418][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:46:11,917][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:46:12,415][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:46:12,914][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:46:13,412][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:46:13,910][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:46:14,409][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:46:14,907][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:46:15,407][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:46:15,909][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:46:16,408][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:46:16,906][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:46:17,404][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21710 tokens. [2026-03-26 04:46:18,038][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 04:46:18,830][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:46:18,833][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:46:18,834][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:46:21,325][__main__][INFO] - Iteration 387 took 1m 16s (9.10% Gen, 87.64% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 55h 20m 5s. Estimated total time: 63h 31m 44s. Time estimates for 10 more iterations: 12m 42s, 100 more iterations: 2h 7m 3s, 500 more iterations: 10h 35m 17s. [2026-03-26 04:46:21,327][__main__][INFO] - Starting iteration 387. [2026-03-26 04:46:21,733][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:46:21,734][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:46:28,614][__main__][INFO] - Number of regex retries in iteration 387: 0 [2026-03-26 04:46:28,614][__main__][INFO] - agents played in iteration 387 are Alice, Bob [2026-03-26 04:46:29,514][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:46:29,558][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:46:29,559][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:46:29,559][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:46:30,165][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:46:30,621][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:46:31,124][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:46:31,623][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:46:32,123][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:46:32,623][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:46:33,123][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:46:33,623][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:46:34,123][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:46:34,621][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:46:35,122][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:46:35,621][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:46:36,120][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:46:36,619][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:46:37,118][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:46:37,617][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:46:38,115][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:46:38,616][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:46:39,116][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:46:39,618][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:46:40,119][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:46:40,620][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:46:41,119][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:46:41,619][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:46:42,120][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:46:42,619][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:46:43,119][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:46:43,619][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:46:44,118][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:46:44,618][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:46:45,117][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:46:45,616][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:46:46,115][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:46:46,614][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:46:47,112][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:46:47,610][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:46:48,109][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:46:48,610][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:46:49,108][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:46:49,608][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:46:50,106][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:46:50,606][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:46:51,105][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:46:51,605][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:46:52,105][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:46:52,604][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:46:53,103][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:46:53,604][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:46:54,104][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:46:54,604][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:46:55,105][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:46:55,606][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:46:56,107][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:46:56,607][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:46:57,107][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:46:57,607][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:46:58,107][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:46:58,607][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:46:59,108][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:46:59,608][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:47:00,107][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:47:00,607][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:47:01,107][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:47:01,607][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:47:02,107][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:47:02,606][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:47:03,106][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:47:03,605][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:47:04,104][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:47:04,604][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:47:05,104][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:47:05,604][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:47:06,103][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:47:06,603][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:47:07,102][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:47:07,601][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:47:08,100][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:47:08,600][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:47:09,099][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:47:09,599][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:47:10,098][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:47:10,599][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:47:11,099][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:47:11,599][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:47:12,100][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:47:12,601][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:47:13,101][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:47:13,602][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:47:14,102][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:47:14,603][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:47:15,102][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:47:15,601][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:47:16,101][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:47:16,601][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:47:17,100][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:47:17,599][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:47:18,099][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:47:18,599][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:47:19,099][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:47:19,599][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:47:20,098][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:47:20,596][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:47:21,094][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:47:21,594][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:47:22,092][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:47:22,590][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:47:23,090][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:47:23,589][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:47:24,089][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:47:24,587][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:47:25,087][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:47:25,586][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:47:26,085][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:47:26,584][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:47:27,084][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:47:27,584][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:47:28,084][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:47:28,583][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:47:29,083][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:47:29,583][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:47:30,084][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:47:30,584][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:47:31,084][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:47:31,584][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:47:32,083][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:47:32,583][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:47:33,082][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:47:33,582][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:47:34,082][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 04:47:34,714][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:47:35,460][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:47:35,462][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:47:35,464][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:47:36,144][__main__][INFO] - Iteration 388 took 1m 14s (9.25% Gen, 89.84% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 47m 41s. Estimated total time: 62h 0m 35s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 1s, 500 more iterations: 10h 20m 5s. [2026-03-26 04:47:36,146][__main__][INFO] - Starting iteration 388. [2026-03-26 04:47:36,552][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:47:36,553][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:47:37,146][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:47:39,364][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:47:42,917][__main__][INFO] - Number of regex retries in iteration 388: 2 [2026-03-26 04:47:42,918][__main__][INFO] - agents played in iteration 388 are Alice, Bob [2026-03-26 04:47:43,800][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:47:43,845][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:47:43,845][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:47:43,846][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:47:44,457][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:47:44,913][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:47:45,421][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:47:45,925][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:47:46,426][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:47:46,927][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:47:47,427][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:47:47,928][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:47:48,430][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:47:48,936][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:47:49,439][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:47:49,942][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:47:50,444][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:47:50,947][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:47:51,450][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:47:51,952][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:47:52,454][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:47:52,952][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:47:53,451][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:47:53,949][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:47:54,447][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:47:54,946][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:47:55,445][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:47:55,944][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:47:56,444][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:47:56,944][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:47:57,443][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:47:57,943][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:47:58,443][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:47:58,942][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:47:59,441][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:47:59,940][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:48:00,439][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:48:00,939][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:48:01,440][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:48:01,941][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:48:02,442][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:48:02,942][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:48:03,442][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:48:03,942][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:48:04,442][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:48:04,942][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:48:05,442][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:48:05,941][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:48:06,441][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:48:06,940][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:48:07,439][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:48:07,939][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:48:08,438][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:48:08,938][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:48:09,437][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:48:09,937][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:48:10,436][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:48:10,934][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:48:11,434][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:48:11,932][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:48:12,430][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:48:12,930][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:48:13,430][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:48:13,930][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:48:14,430][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:48:14,930][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:48:15,430][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:48:15,930][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:48:16,430][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:48:16,931][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:48:17,430][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:48:17,931][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:48:18,434][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:48:18,934][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:48:19,435][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:48:19,936][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:48:20,437][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:48:20,938][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:48:21,441][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:48:21,943][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:48:22,446][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:48:22,948][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:48:23,451][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:48:23,953][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:48:24,455][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:48:24,953][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:48:25,452][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:48:25,951][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:48:26,449][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:48:26,948][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:48:27,447][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:48:27,946][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:48:28,445][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:48:28,944][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:48:29,443][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:48:29,943][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:48:30,442][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:48:30,941][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:48:31,439][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:48:31,939][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:48:32,437][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:48:32,937][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:48:33,439][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:48:33,940][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:48:34,441][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:48:34,941][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:48:35,442][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:48:35,943][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:48:36,444][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:48:36,944][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:48:37,446][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:48:37,946][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:48:38,445][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:48:38,945][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:48:39,444][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:48:39,944][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:48:40,444][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:48:40,944][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:48:41,443][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:48:41,942][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:48:42,441][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:48:42,940][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:48:43,439][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:48:43,940][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:48:44,439][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:48:44,940][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:48:45,439][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:48:45,939][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:48:46,439][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:48:46,940][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:48:47,440][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:48:47,941][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:48:48,444][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:48:49,088][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 04:48:49,823][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:48:49,825][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:48:49,827][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:48:50,605][__main__][INFO] - Iteration 389 took 1m 14s (8.60% Gen, 90.35% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 28m 32s. Estimated total time: 61h 42m 41s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 25s, 500 more iterations: 10h 17m 6s. [2026-03-26 04:48:50,607][__main__][INFO] - Starting iteration 389. [2026-03-26 04:48:51,010][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:48:51,011][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:48:57,593][__main__][INFO] - Number of regex retries in iteration 389: 0 [2026-03-26 04:48:57,594][__main__][INFO] - agents played in iteration 389 are Alice, Bob [2026-03-26 04:48:58,488][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:48:58,533][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:48:58,534][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:48:58,534][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:48:59,144][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:48:59,603][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:49:00,107][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:49:00,607][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:49:01,106][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:49:01,606][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:49:02,106][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:49:02,605][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:49:03,105][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:49:03,606][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:49:04,107][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:49:04,608][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:49:05,108][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:49:05,608][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:49:06,108][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:49:06,608][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:49:07,108][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:49:07,607][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:49:08,106][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:49:08,604][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:49:09,104][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:49:09,604][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:49:10,104][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:49:10,602][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:49:11,101][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:49:11,603][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:49:12,104][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:49:12,605][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:49:13,106][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:49:13,608][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:49:14,108][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:49:14,608][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:49:15,108][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:49:15,612][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:49:16,112][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:49:16,613][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:49:17,112][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:49:17,611][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:49:18,111][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:49:18,611][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:49:19,110][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:49:19,610][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:49:20,109][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:49:20,609][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:49:21,109][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:49:21,608][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:49:22,108][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:49:22,608][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:49:23,107][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:49:23,604][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:49:24,098][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:49:24,593][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:49:25,089][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:49:25,585][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:49:26,082][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:49:26,578][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:49:27,073][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:49:27,573][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:49:28,071][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:49:28,570][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:49:29,069][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:49:29,568][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:49:30,067][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:49:30,568][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:49:31,068][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:49:31,567][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:49:32,067][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:49:32,566][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:49:33,064][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:49:33,565][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:49:34,065][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:49:34,564][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:49:35,064][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:49:35,565][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:49:36,064][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:49:36,564][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:49:37,064][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:49:37,565][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:49:38,067][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:49:38,567][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:49:39,067][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:49:39,565][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:49:40,064][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:49:40,564][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:49:41,063][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:49:41,562][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:49:42,061][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:49:42,560][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:49:43,059][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:49:43,560][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:49:44,061][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:49:44,561][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:49:45,062][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:49:45,562][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:49:46,063][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:49:46,563][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:49:47,064][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:49:47,563][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:49:48,064][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:49:48,566][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:49:49,066][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:49:49,565][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:49:50,064][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:49:50,563][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:49:51,062][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:49:51,563][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:49:52,063][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:49:52,563][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:49:53,061][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:49:53,561][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:49:54,061][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:49:54,561][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:49:55,062][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:49:55,559][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:49:56,055][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:49:56,551][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:49:57,046][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:49:57,541][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:49:58,036][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:49:58,530][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:49:59,025][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:49:59,524][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:50:00,024][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:50:00,523][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:50:01,022][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:50:01,522][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:50:02,021][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:50:02,520][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:50:03,019][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 04:50:03,672][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:50:04,459][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:50:04,462][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:50:04,463][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:50:05,244][__main__][INFO] - Iteration 390 took 1m 14s (8.87% Gen, 90.08% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 36m 21s. Estimated total time: 61h 51m 44s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 43s, 500 more iterations: 10h 18m 37s. [2026-03-26 04:50:05,246][__main__][INFO] - Starting iteration 390. [2026-03-26 04:50:05,656][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:50:05,656][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:50:12,130][__main__][INFO] - Number of regex retries in iteration 390: 0 [2026-03-26 04:50:12,131][__main__][INFO] - agents played in iteration 390 are Alice, Bob [2026-03-26 04:50:13,029][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:50:13,073][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:50:13,074][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:50:13,074][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:50:13,686][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:50:14,143][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:50:14,649][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:50:15,149][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:50:15,648][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:50:16,150][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:50:16,649][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:50:17,149][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:50:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:50:18,151][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:50:18,651][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:50:19,151][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:50:19,651][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:50:20,150][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:50:20,650][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:50:21,150][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:50:21,649][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:50:22,150][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:50:22,650][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:50:23,150][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:50:23,649][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:50:24,149][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:50:24,648][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:50:25,148][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:50:25,648][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:50:26,148][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:50:26,647][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:50:27,147][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:50:27,646][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:50:28,147][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:50:28,646][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:50:29,146][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:50:29,646][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:50:30,146][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:50:30,645][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:50:31,145][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:50:31,644][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:50:32,144][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:50:32,643][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:50:33,143][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:50:33,642][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:50:34,142][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:50:34,641][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:50:35,142][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:50:35,641][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:50:36,141][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:50:36,642][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:50:37,141][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:50:37,640][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:50:38,139][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:50:38,640][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:50:39,137][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:50:39,636][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:50:40,134][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:50:40,631][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:50:41,129][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:50:41,628][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:50:42,127][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:50:42,625][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:50:43,124][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:50:43,623][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:50:44,122][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:50:44,621][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:50:45,120][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:50:45,622][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:50:46,125][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:50:46,626][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:50:47,126][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:50:47,627][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:50:48,129][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:50:48,631][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:50:49,131][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:50:49,631][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:50:50,131][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:50:50,630][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:50:51,130][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:50:51,630][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:50:52,130][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:50:52,630][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:50:53,128][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:50:53,628][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:50:54,128][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:50:54,628][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:50:55,129][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:50:55,630][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:50:56,129][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:50:56,628][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:50:57,127][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:50:57,627][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:50:58,126][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:50:58,625][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:50:59,124][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:50:59,623][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:51:00,122][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:51:00,621][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:51:01,120][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:51:01,618][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:51:02,119][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:51:02,620][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:51:03,120][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:51:03,618][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:51:04,117][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:51:04,615][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:51:05,114][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:51:05,612][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:51:06,111][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:51:06,610][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:51:07,110][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:51:07,609][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:51:08,108][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:51:08,607][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:51:09,106][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:51:09,606][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:51:10,103][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:51:10,602][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:51:11,100][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:51:11,598][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:51:12,095][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:51:12,593][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:51:13,094][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:51:13,593][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:51:14,092][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:51:14,590][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:51:15,090][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:51:15,589][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:51:16,089][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:51:16,588][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:51:17,087][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:51:17,592][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 04:51:18,248][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 04:51:18,997][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:51:19,000][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:51:19,001][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:51:19,934][__main__][INFO] - Iteration 391 took 1m 14s (8.72% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 37m 18s. Estimated total time: 61h 53m 56s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 59s. [2026-03-26 04:51:19,936][__main__][INFO] - Starting iteration 391. [2026-03-26 04:51:20,333][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:51:20,334][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:51:21,389][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:51:27,194][__main__][INFO] - Number of regex retries in iteration 391: 1 [2026-03-26 04:51:27,195][__main__][INFO] - agents played in iteration 391 are Alice, Bob [2026-03-26 04:51:28,069][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:51:28,114][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:51:28,114][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:51:28,115][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:51:28,717][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:51:29,171][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:51:29,673][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:51:30,172][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:51:30,670][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:51:31,169][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:51:31,668][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:51:32,168][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:51:32,667][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:51:33,166][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:51:33,664][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:51:34,163][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:51:34,662][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:51:35,162][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:51:35,661][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:51:36,160][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:51:36,659][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:51:37,156][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:51:37,654][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:51:38,153][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:51:38,650][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:51:39,148][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:51:39,646][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:51:40,144][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:51:40,643][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:51:41,143][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:51:41,643][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:51:42,142][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:51:42,641][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:51:43,141][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:51:43,640][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:51:44,140][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:51:44,639][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:51:45,137][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:51:45,635][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:51:46,133][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:51:46,632][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:51:47,131][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:51:47,630][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:51:48,130][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:51:48,629][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:51:49,128][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:51:49,626][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:51:50,125][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:51:50,624][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:51:51,123][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:51:51,622][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:51:52,121][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:51:52,619][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:51:53,117][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:51:53,616][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:51:54,114][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:51:54,613][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:51:55,112][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:51:55,610][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:51:56,109][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:51:56,607][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:51:57,106][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:51:57,604][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:51:58,103][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:51:58,601][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:51:59,101][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:51:59,600][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:52:00,100][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:52:00,598][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:52:01,097][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:52:01,596][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:52:02,094][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:52:02,593][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:52:03,092][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:52:03,591][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:52:04,091][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:52:04,590][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:52:05,088][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:52:05,586][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:52:06,085][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:52:06,582][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:52:07,081][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:52:07,582][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:52:08,080][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:52:08,580][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:52:09,078][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:52:09,577][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:52:10,075][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:52:10,573][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:52:11,071][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:52:11,569][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:52:12,068][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:52:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:52:13,066][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:52:13,565][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:52:14,064][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:52:14,562][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:52:15,063][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:52:15,565][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:52:16,066][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:52:16,565][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:52:17,064][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:52:17,564][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:52:18,064][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:52:18,565][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:52:19,064][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:52:19,563][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:52:20,061][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:52:20,558][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:52:21,056][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:52:21,555][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:52:22,052][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:52:22,551][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:52:23,049][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:52:23,547][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:52:24,046][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:52:24,544][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:52:25,044][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:52:25,543][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:52:26,042][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:52:26,541][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:52:27,040][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:52:27,539][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:52:28,039][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:52:28,537][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:52:29,036][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:52:29,534][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:52:30,032][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:52:30,530][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:52:31,029][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:52:31,528][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:52:32,027][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:52:32,525][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21727 tokens. [2026-03-26 04:52:33,162][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:52:34,112][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:52:34,114][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:52:34,116][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:52:34,982][__main__][INFO] - Iteration 392 took 1m 14s (9.19% Gen, 89.65% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 54m 33s. Estimated total time: 62h 12m 26s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 4s. [2026-03-26 04:52:34,984][__main__][INFO] - Starting iteration 392. [2026-03-26 04:52:35,387][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:52:35,387][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:52:36,197][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 04:52:41,951][__main__][INFO] - Number of regex retries in iteration 392: 1 [2026-03-26 04:52:41,952][__main__][INFO] - agents played in iteration 392 are Alice, Bob [2026-03-26 04:52:42,850][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:52:42,894][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:52:42,895][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:52:42,895][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:52:43,499][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:52:43,953][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:52:44,455][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:52:44,954][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:52:45,452][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:52:45,951][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:52:46,449][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:52:46,947][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:52:47,445][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:52:47,948][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:52:48,449][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:52:48,949][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:52:49,449][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:52:49,949][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:52:50,447][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:52:50,947][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:52:51,446][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:52:51,947][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:52:52,447][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:52:52,947][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:52:53,446][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:52:53,946][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:52:54,446][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:52:54,945][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:52:55,446][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:52:55,946][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:52:56,445][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:52:56,945][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:52:57,444][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:52:57,945][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:52:58,444][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:52:58,943][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:52:59,443][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:52:59,942][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:53:00,440][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:53:00,939][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:53:01,437][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:53:01,934][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:53:02,431][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:53:02,928][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:53:03,426][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:53:03,926][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:53:04,424][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:53:04,923][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:53:05,423][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:53:05,921][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:53:06,421][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:53:06,919][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:53:07,419][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:53:07,919][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:53:08,418][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:53:08,918][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:53:09,417][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:53:09,915][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:53:10,413][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:53:10,912][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:53:11,411][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:53:11,909][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:53:12,408][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:53:12,907][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:53:13,407][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:53:13,906][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:53:14,406][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:53:14,905][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:53:15,404][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:53:15,902][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:53:16,401][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:53:16,900][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:53:17,398][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:53:17,897][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:53:18,396][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:53:18,894][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:53:19,392][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:53:19,891][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:53:20,391][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:53:20,891][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:53:21,390][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:53:21,889][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:53:22,388][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:53:22,888][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:53:23,388][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:53:23,887][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:53:24,386][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:53:24,885][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:53:25,384][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:53:25,883][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:53:26,381][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:53:26,882][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:53:27,381][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:53:27,881][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:53:28,381][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:53:28,881][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:53:29,382][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:53:29,881][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:53:30,382][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:53:30,882][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:53:31,382][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:53:31,881][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:53:32,378][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:53:32,875][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:53:33,372][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:53:33,869][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:53:34,368][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:53:34,866][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:53:35,364][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:53:35,863][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:53:36,362][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:53:36,862][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:53:37,362][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:53:37,861][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:53:38,362][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:53:38,862][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:53:39,361][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:53:39,861][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:53:40,360][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:53:40,860][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:53:41,361][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:53:41,861][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:53:42,360][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:53:42,860][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:53:43,358][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:53:43,856][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:53:44,354][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:53:44,852][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:53:45,351][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:53:45,852][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:53:46,350][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:53:46,851][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:53:47,350][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 04:53:47,978][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-26 04:53:48,723][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:53:48,726][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:53:48,727][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:53:49,375][__main__][INFO] - Iteration 393 took 1m 13s (8.87% Gen, 90.25% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 20m 20s. Estimated total time: 61h 39m 27s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 18s, 500 more iterations: 10h 16m 34s. [2026-03-26 04:53:49,378][__main__][INFO] - Starting iteration 393. [2026-03-26 04:53:49,777][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:53:49,778][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:53:56,470][__main__][INFO] - Number of regex retries in iteration 393: 0 [2026-03-26 04:53:56,471][__main__][INFO] - agents played in iteration 393 are Alice, Bob [2026-03-26 04:53:57,379][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:53:57,423][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:53:57,424][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:53:57,424][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:53:58,038][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:53:58,493][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:53:58,996][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:53:59,495][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:53:59,995][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:54:00,494][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:54:00,993][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:54:01,492][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:54:01,992][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:54:02,490][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:54:02,988][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:54:03,488][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:54:03,987][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:54:04,486][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:54:04,986][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:54:05,484][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:54:05,983][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:54:06,485][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:54:06,985][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:54:07,485][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:54:07,985][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:54:08,485][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:54:08,985][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:54:09,485][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:54:09,985][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:54:10,485][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:54:10,984][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:54:11,483][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:54:11,983][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:54:12,482][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:54:12,981][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:54:13,481][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:54:13,981][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:54:14,483][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:54:14,983][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:54:15,482][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:54:15,981][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:54:16,481][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:54:16,982][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:54:17,482][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:54:17,983][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:54:18,487][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:54:18,989][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:54:19,489][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:54:19,990][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:54:20,491][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:54:20,992][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:54:21,492][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:54:21,994][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:54:22,493][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:54:22,992][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:54:23,491][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:54:23,990][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:54:24,489][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:54:24,990][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:54:25,490][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:54:25,990][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:54:26,489][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:54:26,988][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:54:27,487][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:54:27,987][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:54:28,487][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:54:28,987][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:54:29,486][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:54:29,986][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:54:30,485][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:54:30,985][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:54:31,484][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:54:31,983][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:54:32,482][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:54:32,982][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:54:33,482][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:54:33,982][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:54:34,480][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:54:34,979][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:54:35,478][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:54:35,976][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:54:36,474][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:54:36,973][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:54:37,471][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:54:37,969][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:54:38,469][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:54:38,969][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:54:39,468][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:54:39,968][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:54:40,467][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:54:40,967][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:54:41,468][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:54:41,968][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:54:42,467][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:54:42,968][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:54:43,468][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:54:43,967][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:54:44,467][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:54:44,966][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:54:45,466][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:54:45,966][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:54:46,466][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:54:46,964][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:54:47,464][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:54:47,963][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:54:48,465][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:54:48,965][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:54:49,464][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:54:49,963][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:54:50,464][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:54:50,967][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:54:51,467][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:54:51,968][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:54:52,468][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:54:52,969][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:54:53,470][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:54:53,971][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:54:54,471][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:54:54,970][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:54:55,469][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:54:55,969][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:54:56,468][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:54:56,968][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:54:57,468][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:54:57,968][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:54:58,467][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:54:58,966][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:54:59,465][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:54:59,963][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:55:00,462][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:55:00,963][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:55:01,462][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:55:01,961][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 04:55:02,613][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 04:55:03,396][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:55:03,398][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:55:03,400][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:55:04,115][__main__][INFO] - Iteration 394 took 1m 14s (9.00% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 36m 34s. Estimated total time: 61h 56m 56s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 53s, 500 more iterations: 10h 19m 29s. [2026-03-26 04:55:04,117][__main__][INFO] - Starting iteration 394. [2026-03-26 04:55:04,517][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:55:04,518][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:55:10,628][__main__][INFO] - Number of regex retries in iteration 394: 0 [2026-03-26 04:55:10,628][__main__][INFO] - agents played in iteration 394 are Alice, Bob [2026-03-26 04:55:11,502][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:55:11,826][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:55:11,827][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:55:11,828][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:55:12,436][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:55:12,895][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:55:13,398][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:55:13,896][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:55:14,395][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:55:14,893][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:55:15,394][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:55:15,895][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:55:16,394][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:55:16,895][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:55:17,393][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:55:17,893][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:55:18,392][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:55:18,892][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:55:19,392][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:55:19,891][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:55:20,391][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:55:20,891][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:55:21,391][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:55:21,891][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:55:22,390][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:55:22,891][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:55:23,391][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:55:23,890][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:55:24,390][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:55:24,888][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:55:25,386][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:55:25,885][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:55:26,382][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:55:26,880][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:55:27,379][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:55:27,875][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:55:28,372][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:55:28,870][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:55:29,369][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:55:29,868][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:55:30,368][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:55:30,867][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:55:31,366][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:55:31,865][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:55:32,364][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:55:32,863][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:55:33,362][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:55:33,861][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:55:34,361][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:55:34,859][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:55:35,358][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:55:35,857][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:55:36,355][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:55:36,873][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:55:37,378][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:55:37,878][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:55:38,379][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:55:38,879][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:55:39,378][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:55:39,880][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:55:40,382][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:55:40,881][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:55:41,382][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:55:41,882][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:55:42,383][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:55:42,885][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:55:43,385][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:55:43,887][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:55:44,387][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:55:44,885][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:55:45,383][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:55:45,883][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:55:46,382][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:55:46,882][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:55:47,382][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:55:47,882][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:55:48,382][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:55:48,883][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:55:49,383][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:55:49,884][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:55:50,383][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:55:50,882][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:55:51,381][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:55:51,882][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:55:52,380][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:55:52,881][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:55:53,381][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:55:53,881][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:55:54,382][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:55:54,881][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:55:55,382][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:55:55,881][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:55:56,379][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:55:56,878][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:55:57,375][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:55:57,873][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:55:58,370][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:55:58,868][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:55:59,366][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:55:59,863][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:56:00,360][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:56:00,859][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:56:01,356][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:56:01,855][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:56:02,354][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:56:02,852][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:56:03,351][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:56:03,850][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:56:04,349][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:56:04,847][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:56:05,346][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:56:05,846][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:56:06,344][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:56:06,844][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:56:07,342][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:56:07,841][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:56:08,340][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:56:08,840][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:56:09,339][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:56:09,841][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:56:10,342][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:56:10,842][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:56:11,345][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:56:11,846][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:56:12,347][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:56:12,846][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:56:13,347][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:56:13,848][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:56:14,348][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:56:14,848][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:56:15,347][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:56:15,849][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:56:16,349][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21728 tokens. [2026-03-26 04:56:16,995][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.31%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 04:56:17,771][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:56:17,773][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:56:17,775][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:56:18,507][__main__][INFO] - Iteration 395 took 1m 13s (8.26% Gen, 90.75% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 17m 53s. Estimated total time: 61h 39m 29s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 18s, 500 more iterations: 10h 16m 34s. [2026-03-26 04:56:18,509][__main__][INFO] - Starting iteration 395. [2026-03-26 04:56:18,908][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:56:18,909][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:56:25,913][__main__][INFO] - Number of regex retries in iteration 395: 0 [2026-03-26 04:56:25,914][__main__][INFO] - agents played in iteration 395 are Alice, Bob [2026-03-26 04:56:26,810][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:56:26,856][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:56:26,856][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:56:26,857][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:56:27,463][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:56:27,921][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:56:28,424][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:56:28,924][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:56:29,425][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:56:29,926][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:56:30,427][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:56:30,927][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:56:31,432][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:56:31,932][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:56:32,431][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:56:32,930][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:56:33,431][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:56:33,930][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:56:34,430][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:56:34,929][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:56:35,428][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:56:35,934][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:56:36,436][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:56:36,937][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:56:37,438][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:56:37,939][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:56:38,440][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:56:38,940][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:56:39,440][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:56:39,942][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:56:40,449][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:56:40,949][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:56:41,449][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:56:41,949][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:56:42,448][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:56:42,948][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:56:43,449][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:56:43,949][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:56:44,446][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:56:44,943][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:56:45,444][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:56:45,943][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:56:46,440][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:56:46,936][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:56:47,431][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:56:47,932][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:56:48,434][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:56:48,935][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:56:49,434][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:56:49,934][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:56:50,433][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:56:50,933][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:56:51,432][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:56:51,932][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:56:52,431][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:56:52,931][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:56:53,430][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:56:53,930][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:56:54,429][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:56:54,928][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:56:55,428][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:56:55,926][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:56:56,425][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:56:56,922][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:56:57,420][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:56:57,918][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:56:58,416][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:56:58,914][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:56:59,412][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:56:59,911][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:57:00,410][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:57:00,909][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:57:01,408][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:57:01,907][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:57:02,406][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:57:02,906][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:57:03,405][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:57:03,904][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:57:04,403][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:57:04,901][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:57:05,399][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:57:05,897][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:57:06,395][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:57:06,894][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:57:07,393][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:57:07,893][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:57:08,392][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:57:08,891][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:57:09,390][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:57:09,890][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:57:10,390][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:57:10,890][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:57:11,389][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:57:11,888][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:57:12,386][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:57:12,884][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:57:13,383][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:57:13,883][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:57:14,381][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:57:14,880][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:57:15,379][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:57:15,877][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:57:16,374][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:57:16,871][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:57:17,368][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:57:17,867][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:57:18,366][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:57:18,866][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:57:19,363][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:57:19,862][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:57:20,362][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:57:20,860][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:57:21,358][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:57:21,857][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:57:22,355][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:57:22,854][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:57:23,353][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:57:23,852][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:57:24,351][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:57:24,849][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:57:25,348][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:57:25,846][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:57:26,346][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:57:26,844][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:57:27,343][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:57:27,841][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:57:28,340][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:57:28,838][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:57:29,336][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:57:29,834][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:57:30,331][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:57:30,829][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:57:31,327][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21607 tokens. [2026-03-26 04:57:31,968][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 04:57:32,747][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:57:32,749][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:57:32,751][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:57:33,472][__main__][INFO] - Iteration 396 took 1m 14s (9.39% Gen, 89.64% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 45m 23s. Estimated total time: 62h 8m 14s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 16s, 500 more iterations: 10h 21m 22s. [2026-03-26 04:57:33,475][__main__][INFO] - Starting iteration 396. [2026-03-26 04:57:33,888][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:57:33,889][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:57:40,111][__main__][INFO] - Number of regex retries in iteration 396: 0 [2026-03-26 04:57:40,112][__main__][INFO] - agents played in iteration 396 are Alice, Bob [2026-03-26 04:57:41,008][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:57:41,053][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:57:41,054][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:57:41,054][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:57:41,656][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:57:42,114][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:57:42,620][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:57:43,122][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:57:43,624][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:57:44,127][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:57:44,630][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:57:45,134][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:57:45,635][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:57:46,135][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:57:46,636][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:57:47,135][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:57:47,640][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:57:48,142][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:57:48,644][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:57:49,145][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:57:49,644][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:57:50,146][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:57:50,647][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:57:51,148][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:57:51,649][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:57:52,155][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:57:52,656][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:57:53,157][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:57:53,659][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:57:54,162][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:57:54,664][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:57:55,167][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:57:55,670][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:57:56,172][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:57:56,679][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:57:57,184][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:57:57,688][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:57:58,191][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:57:58,694][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:57:59,195][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:57:59,697][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:58:00,200][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:58:00,701][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:58:01,210][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:58:01,713][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:58:02,216][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:58:02,719][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:58:03,220][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:58:03,722][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:58:04,226][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:58:04,728][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:58:05,230][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:58:05,741][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:58:06,241][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:58:06,742][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:58:07,243][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:58:07,742][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:58:08,243][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:58:08,743][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:58:09,242][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:58:09,742][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:58:10,246][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:58:10,747][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:58:11,247][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:58:11,747][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:58:12,247][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:58:12,747][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:58:13,248][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:58:13,749][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:58:14,251][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:58:14,753][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:58:15,255][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:58:15,759][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:58:16,264][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:58:16,766][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:58:17,267][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:58:17,769][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:58:18,270][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:58:18,770][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:58:19,270][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:58:19,770][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:58:20,269][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:58:20,769][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:58:21,268][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:58:21,767][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:58:22,268][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:58:22,768][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:58:23,268][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:58:23,768][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:58:24,268][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:58:24,768][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:58:25,269][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:58:25,769][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:58:26,270][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:58:26,771][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:58:27,272][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:58:27,773][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:58:28,273][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:58:28,773][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:58:29,273][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:58:29,774][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:58:30,274][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:58:30,775][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:58:31,275][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:58:31,775][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:58:32,277][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:58:32,777][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:58:33,277][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:58:33,777][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:58:34,277][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:58:34,777][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:58:35,276][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:58:35,775][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:58:36,275][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:58:36,774][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:58:37,274][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:58:37,774][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:58:38,272][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:58:38,770][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:58:39,268][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:58:39,767][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:58:40,265][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:58:40,764][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:58:41,262][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:58:41,761][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:58:42,259][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:58:42,758][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:58:43,256][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:58:43,754][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:58:44,252][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:58:44,751][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 04:58:45,250][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 04:58:45,749][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 04:58:46,388][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 04:58:47,114][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 04:58:47,116][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 04:58:47,118][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 04:58:47,857][__main__][INFO] - Iteration 397 took 1m 13s (8.41% Gen, 90.59% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 14m 22s. Estimated total time: 61h 38m 28s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 16s, 500 more iterations: 10h 16m 24s. [2026-03-26 04:58:47,859][__main__][INFO] - Starting iteration 397. [2026-03-26 04:58:48,261][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 04:58:48,262][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 04:58:55,459][__main__][INFO] - Number of regex retries in iteration 397: 0 [2026-03-26 04:58:55,459][__main__][INFO] - agents played in iteration 397 are Alice, Bob [2026-03-26 04:58:56,372][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:58:56,417][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 04:58:56,417][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 04:58:56,418][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 04:58:57,029][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 04:58:57,484][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 04:58:57,988][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 04:58:58,489][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 04:58:58,989][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 04:58:59,489][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 04:58:59,989][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 04:59:00,490][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 04:59:00,989][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 04:59:01,489][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 04:59:01,988][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 04:59:02,486][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 04:59:02,984][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 04:59:03,482][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 04:59:03,979][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 04:59:04,477][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 04:59:04,974][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 04:59:05,474][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 04:59:05,973][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 04:59:06,472][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 04:59:06,972][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 04:59:07,471][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 04:59:07,971][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 04:59:08,470][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 04:59:08,970][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 04:59:09,469][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 04:59:09,969][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 04:59:10,468][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 04:59:10,965][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 04:59:11,462][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 04:59:11,960][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 04:59:12,457][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 04:59:12,954][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 04:59:13,453][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 04:59:13,951][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 04:59:14,449][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 04:59:14,948][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 04:59:15,446][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 04:59:15,944][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 04:59:16,442][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 04:59:16,941][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 04:59:17,440][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 04:59:17,939][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 04:59:18,440][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 04:59:18,940][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 04:59:19,438][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 04:59:19,936][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 04:59:20,434][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 04:59:20,933][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 04:59:21,432][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 04:59:21,931][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 04:59:22,430][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 04:59:22,929][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 04:59:23,428][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 04:59:23,926][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 04:59:24,424][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 04:59:24,923][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 04:59:25,423][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 04:59:25,922][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 04:59:26,421][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 04:59:26,920][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 04:59:27,417][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 04:59:27,915][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 04:59:28,412][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 04:59:28,910][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 04:59:29,410][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 04:59:29,909][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 04:59:30,409][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 04:59:30,909][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 04:59:31,408][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 04:59:31,907][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 04:59:32,406][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 04:59:32,905][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 04:59:33,404][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 04:59:33,902][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 04:59:34,400][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 04:59:34,898][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 04:59:35,395][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 04:59:35,893][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 04:59:36,390][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 04:59:36,888][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 04:59:37,386][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 04:59:37,885][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 04:59:38,384][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 04:59:38,883][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 04:59:39,383][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 04:59:39,881][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 04:59:40,379][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 04:59:40,877][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 04:59:41,374][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 04:59:41,872][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 04:59:42,371][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 04:59:42,870][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 04:59:43,368][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 04:59:43,866][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 04:59:44,364][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 04:59:44,868][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 04:59:45,370][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 04:59:45,871][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 04:59:46,371][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 04:59:46,870][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 04:59:47,369][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 04:59:47,869][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 04:59:48,369][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 04:59:48,870][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 04:59:49,371][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 04:59:49,872][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 04:59:50,372][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 04:59:50,872][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 04:59:51,372][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 04:59:51,872][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 04:59:52,373][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 04:59:52,873][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 04:59:53,374][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 04:59:53,879][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 04:59:54,379][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 04:59:54,879][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 04:59:55,380][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 04:59:55,881][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 04:59:56,382][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 04:59:56,882][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 04:59:57,383][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 04:59:57,884][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 04:59:58,388][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 04:59:58,890][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 04:59:59,392][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 04:59:59,893][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:00:00,393][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:00:00,893][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 05:00:01,540][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 05:00:02,290][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:00:02,293][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:00:02,294][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:00:03,036][__main__][INFO] - Iteration 398 took 1m 14s (9.63% Gen, 89.38% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 53m 27s. Estimated total time: 62h 18m 48s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 37s, 500 more iterations: 10h 23m 8s. [2026-03-26 05:00:03,038][__main__][INFO] - Starting iteration 398. [2026-03-26 05:00:03,436][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 05:00:03,437][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:00:09,997][__main__][INFO] - Number of regex retries in iteration 398: 0 [2026-03-26 05:00:09,998][__main__][INFO] - agents played in iteration 398 are Alice, Bob [2026-03-26 05:00:10,883][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:00:10,927][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:00:10,928][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:00:10,929][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:00:11,532][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:00:11,988][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:00:12,494][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:00:12,993][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:00:13,492][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:00:13,991][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:00:14,491][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:00:14,990][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:00:15,490][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:00:15,988][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:00:16,487][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:00:16,985][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:00:17,485][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:00:17,985][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:00:18,486][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:00:18,985][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:00:19,484][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:00:19,985][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:00:20,486][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:00:20,986][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:00:21,487][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:00:21,989][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:00:22,490][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:00:22,990][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:00:23,491][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:00:23,990][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:00:24,489][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:00:24,989][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:00:25,490][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:00:25,990][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:00:26,490][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:00:26,990][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:00:27,489][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:00:27,991][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:00:28,491][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:00:28,992][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:00:29,492][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:00:29,992][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:00:30,492][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:00:30,992][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:00:31,491][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:00:31,992][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:00:32,492][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:00:32,992][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:00:33,491][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:00:33,991][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:00:34,491][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:00:34,990][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:00:35,490][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:00:35,989][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:00:36,488][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:00:36,988][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:00:37,487][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:00:37,986][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:00:38,486][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:00:38,985][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:00:39,484][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:00:39,984][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:00:40,483][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:00:40,983][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:00:41,483][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:00:41,982][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:00:42,481][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:00:42,981][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:00:43,481][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:00:43,980][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:00:44,479][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:00:44,978][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:00:45,477][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:00:45,975][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:00:46,474][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:00:46,972][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:00:47,473][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:00:47,973][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:00:48,471][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:00:48,970][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:00:49,469][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:00:49,969][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:00:50,466][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:00:50,966][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:00:51,464][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:00:51,964][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:00:52,464][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:00:52,964][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:00:53,464][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:00:53,965][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:00:54,465][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:00:54,965][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:00:55,466][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:00:55,965][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:00:56,464][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:00:56,964][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:00:57,463][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:00:57,965][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:00:58,465][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:00:58,964][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:00:59,463][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:00:59,963][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:01:00,464][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:01:00,964][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:01:01,465][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:01:01,965][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:01:02,465][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:01:02,964][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:01:03,465][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:01:03,965][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:01:04,464][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:01:04,965][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:01:05,464][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:01:05,964][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:01:06,463][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:01:06,963][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:01:07,463][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:01:07,961][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:01:08,460][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:01:08,957][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:01:09,455][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:01:09,955][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:01:10,456][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:01:10,955][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:01:11,455][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:01:11,953][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:01:12,452][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:01:12,951][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:01:13,450][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:01:13,950][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:01:14,449][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:01:14,955][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:01:15,457][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 05:01:16,099][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 05:01:16,893][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:01:16,896][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:01:16,898][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:01:17,668][__main__][INFO] - Iteration 399 took 1m 14s (8.84% Gen, 90.12% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 25m 0s. Estimated total time: 61h 51m 35s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 43s, 500 more iterations: 10h 18m 35s. [2026-03-26 05:01:17,670][__main__][INFO] - Starting iteration 399. [2026-03-26 05:01:18,078][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 05:01:18,079][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:01:24,697][__main__][INFO] - Number of regex retries in iteration 399: 0 [2026-03-26 05:01:24,697][__main__][INFO] - agents played in iteration 399 are Alice, Bob [2026-03-26 05:01:25,593][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:01:25,639][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:01:25,640][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:01:25,640][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:01:26,246][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:01:26,704][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:01:27,210][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:01:27,711][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:01:28,212][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:01:28,713][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:01:29,214][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:01:29,714][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:01:30,221][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:01:30,722][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:01:31,223][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:01:31,724][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:01:32,225][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:01:32,724][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:01:33,225][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:01:33,725][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:01:34,225][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:01:34,733][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:01:35,234][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:01:35,735][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:01:36,235][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:01:36,736][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:01:37,236][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:01:37,736][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:01:38,237][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:01:38,737][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:01:39,244][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:01:39,745][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:01:40,247][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:01:40,748][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:01:41,248][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:01:41,748][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:01:42,248][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:01:42,749][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:01:43,250][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:01:43,752][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:01:44,252][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:01:44,752][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:01:45,251][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:01:45,751][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:01:46,250][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:01:46,750][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:01:47,251][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:01:47,752][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:01:48,254][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:01:48,756][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:01:49,256][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:01:49,756][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:01:50,256][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:01:50,754][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:01:51,253][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:01:51,751][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:01:52,249][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:01:52,748][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:01:53,245][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:01:53,743][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:01:54,244][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:01:54,744][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:01:55,246][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:01:55,745][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:01:56,246][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:01:56,747][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:01:57,248][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:01:57,748][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:01:58,249][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:01:58,749][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:01:59,250][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:01:59,750][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:02:00,249][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:02:00,750][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:02:01,251][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:02:01,751][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:02:02,251][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:02:02,750][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:02:03,249][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:02:03,748][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:02:04,248][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:02:04,746][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:02:05,245][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:02:05,744][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:02:06,243][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:02:06,742][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:02:07,241][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:02:07,739][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:02:08,237][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:02:08,735][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:02:09,233][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:02:09,732][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:02:10,231][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:02:10,729][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:02:11,228][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:02:11,727][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:02:12,227][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:02:12,727][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:02:13,227][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:02:13,727][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:02:14,225][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:02:14,725][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:02:15,225][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:02:15,724][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:02:16,224][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:02:16,725][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:02:17,225][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:02:17,724][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:02:18,226][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:02:18,726][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:02:19,226][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:02:19,726][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:02:20,226][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:02:20,725][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:02:21,225][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:02:21,724][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:02:22,224][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:02:22,721][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:02:23,219][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:02:23,717][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:02:24,214][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:02:24,712][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:02:25,210][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:02:25,709][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:02:26,206][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:02:26,706][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:02:27,207][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:02:27,707][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:02:28,207][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:02:28,707][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:02:29,208][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:02:29,708][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:02:30,208][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 05:02:30,849][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:02:31,577][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:02:31,579][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:02:31,581][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:02:32,374][__main__][INFO] - Iteration 400 took 1m 14s (8.91% Gen, 90.02% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 26m 59s. Estimated total time: 61h 54m 49s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 49s, 500 more iterations: 10h 19m 8s. [2026-03-26 05:02:32,376][__main__][INFO] - Starting iteration 400. [2026-03-26 05:02:32,774][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 7 and human policies 1. [2026-03-26 05:02:32,775][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:02:40,101][__main__][INFO] - Number of regex retries in iteration 400: 0 [2026-03-26 05:02:40,102][__main__][INFO] - agents played in iteration 400 are Alice, Bob [2026-03-26 05:02:41,000][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:02:41,045][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:02:41,046][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:02:41,046][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:02:41,646][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:02:42,100][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:02:42,604][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:02:43,110][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:02:43,611][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:02:44,111][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:02:44,612][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:02:45,111][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:02:45,616][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:02:46,121][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:02:46,625][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:02:47,127][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:02:47,629][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:02:48,132][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:02:48,636][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:02:49,138][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:02:49,640][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:02:50,140][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:02:50,642][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:02:51,142][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:02:51,643][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:02:52,150][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:02:52,652][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:02:53,153][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:02:53,654][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:02:54,154][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:02:54,655][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:02:55,156][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:02:55,656][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:02:56,157][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:02:56,659][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:02:57,160][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:02:57,662][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:02:58,165][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:02:58,666][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:02:59,167][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:02:59,668][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:03:00,170][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:03:00,671][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:03:01,179][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:03:01,681][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:03:02,181][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:03:02,682][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:03:03,183][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:03:03,685][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:03:04,186][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:03:04,688][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:03:05,189][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:03:05,697][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:03:06,198][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:03:06,699][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:03:07,199][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:03:07,701][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:03:08,201][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:03:08,700][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:03:09,199][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:03:09,700][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:03:10,204][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:03:10,705][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:03:11,206][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:03:11,705][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:03:12,205][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:03:12,705][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:03:13,206][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:03:13,706][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:03:14,206][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:03:14,706][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:03:15,206][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:03:15,705][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:03:16,205][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:03:16,705][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:03:17,205][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:03:17,706][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:03:18,209][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:03:18,711][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:03:19,212][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:03:19,714][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:03:20,215][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:03:20,716][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:03:21,217][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:03:21,718][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:03:22,218][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:03:22,718][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:03:23,218][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:03:23,718][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:03:24,219][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:03:24,718][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:03:25,218][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:03:25,717][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:03:26,216][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:03:26,715][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:03:27,213][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:03:27,713][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:03:28,213][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:03:28,713][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:03:29,211][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:03:29,711][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:03:30,211][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:03:30,710][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:03:31,210][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:03:31,710][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:03:32,210][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:03:32,710][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:03:33,210][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:03:33,710][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:03:34,208][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:03:34,706][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:03:35,205][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:03:35,703][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:03:36,201][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:03:36,699][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:03:37,197][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:03:37,695][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:03:38,193][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:03:38,692][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:03:39,190][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:03:39,689][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:03:40,187][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:03:40,685][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:03:41,184][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:03:41,682][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:03:42,181][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:03:42,680][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:03:43,178][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:03:43,677][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:03:44,175][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:03:44,673][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:03:45,171][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:03:45,670][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 05:03:46,307][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:03:47,157][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:03:47,160][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:03:47,161][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:03:50,325][__main__][INFO] - Iteration 401 took 1m 17s (9.45% Gen, 86.47% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 56h 8m 25s. Estimated total time: 64h 37m 34s. Time estimates for 10 more iterations: 12m 55s, 100 more iterations: 2h 9m 15s, 500 more iterations: 10h 46m 15s. [2026-03-26 05:03:50,327][__main__][INFO] - Starting iteration 401. [2026-03-26 05:03:50,727][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:03:50,728][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:03:51,349][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:03:57,307][__main__][INFO] - Number of regex retries in iteration 401: 1 [2026-03-26 05:03:57,308][__main__][INFO] - agents played in iteration 401 are Alice, Bob [2026-03-26 05:03:58,189][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:03:58,234][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:03:58,234][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:03:58,235][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:03:58,841][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:03:59,295][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:03:59,798][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:04:00,296][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:04:00,795][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:04:01,294][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:04:01,794][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:04:02,294][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:04:02,792][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:04:03,291][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:04:03,789][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:04:04,288][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:04:04,787][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:04:05,286][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:04:05,785][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:04:06,283][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:04:06,782][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:04:07,279][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:04:07,776][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:04:08,273][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:04:08,771][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:04:09,269][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:04:09,767][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:04:10,265][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:04:10,764][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:04:11,263][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:04:11,761][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:04:12,259][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:04:12,757][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:04:13,254][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:04:13,753][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:04:14,252][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:04:14,750][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:04:15,253][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:04:15,753][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:04:16,253][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:04:16,752][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:04:17,252][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:04:17,753][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:04:18,255][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:04:18,756][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:04:19,254][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:04:19,752][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:04:20,250][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:04:20,749][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:04:21,248][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:04:21,746][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:04:22,245][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:04:22,743][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:04:23,241][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:04:23,740][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:04:24,238][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:04:24,735][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:04:25,234][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:04:25,732][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:04:26,230][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:04:26,728][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:04:27,225][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:04:27,723][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:04:28,222][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:04:28,719][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:04:29,217][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:04:29,715][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:04:30,213][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:04:30,712][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:04:31,210][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:04:31,707][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:04:32,205][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:04:32,703][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:04:33,200][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:04:33,698][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:04:34,195][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:04:34,694][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:04:35,192][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:04:35,691][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:04:36,190][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:04:36,690][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:04:37,189][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:04:37,687][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:04:38,187][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:04:38,686][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:04:39,185][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:04:39,685][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:04:40,183][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:04:40,681][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:04:41,179][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:04:41,676][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:04:42,174][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:04:42,674][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:04:43,173][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:04:43,671][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:04:44,169][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:04:44,667][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:04:45,167][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:04:45,665][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:04:46,166][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:04:46,666][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:04:47,166][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:04:47,666][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:04:48,169][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:04:48,671][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:04:49,172][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:04:49,673][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:04:50,180][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:04:50,682][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:04:51,181][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:04:51,680][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:04:52,180][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:04:52,680][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:04:53,180][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:04:53,680][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:04:54,179][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:04:54,685][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:04:55,188][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:04:55,689][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:04:56,189][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:04:56,690][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:04:57,189][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:04:57,689][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:04:58,190][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:04:58,691][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:04:59,199][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:04:59,701][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:05:00,202][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:05:00,702][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:05:01,202][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:05:01,704][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:05:02,205][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:05:02,706][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 05:05:03,354][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:05:04,134][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:05:04,137][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:05:04,138][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:05:05,337][__main__][INFO] - Iteration 402 took 1m 14s (8.82% Gen, 89.57% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 40m 7s. Estimated total time: 62h 10m 30s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 21s, 500 more iterations: 10h 21m 45s. [2026-03-26 05:05:05,339][__main__][INFO] - Starting iteration 402. [2026-03-26 05:05:05,747][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:05:05,748][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:05:06,357][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:05:08,590][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:05:12,702][__main__][INFO] - Number of regex retries in iteration 402: 2 [2026-03-26 05:05:12,703][__main__][INFO] - agents played in iteration 402 are Alice, Bob [2026-03-26 05:05:13,594][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:05:13,639][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:05:13,639][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:05:13,640][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:05:14,245][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:05:14,701][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:05:15,207][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:05:15,708][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:05:16,208][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:05:16,709][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:05:17,209][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:05:17,708][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:05:18,209][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:05:18,710][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:05:19,210][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:05:19,711][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:05:20,211][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:05:20,711][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:05:21,210][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:05:21,711][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:05:22,211][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:05:22,716][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:05:23,218][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:05:23,719][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:05:24,219][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:05:24,719][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:05:25,219][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:05:25,719][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:05:26,219][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:05:26,720][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:05:27,220][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:05:27,720][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:05:28,220][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:05:28,720][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:05:29,220][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:05:29,721][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:05:30,221][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:05:30,719][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:05:31,217][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:05:31,715][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:05:32,213][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:05:32,712][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:05:33,211][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:05:33,710][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:05:34,209][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:05:34,709][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:05:35,208][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:05:35,708][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:05:36,207][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:05:36,708][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:05:37,207][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:05:37,706][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:05:38,206][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:05:38,705][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:05:39,204][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:05:39,704][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:05:40,205][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:05:40,706][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:05:41,206][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:05:41,706][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:05:42,206][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:05:42,705][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:05:43,206][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:05:43,706][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:05:44,207][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:05:44,708][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:05:45,208][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:05:45,708][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:05:46,209][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:05:46,709][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:05:47,209][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:05:47,709][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:05:48,211][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:05:48,712][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:05:49,212][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:05:49,712][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:05:50,212][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:05:50,712][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:05:51,212][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:05:51,712][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:05:52,213][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:05:52,712][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:05:53,212][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:05:53,712][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:05:54,212][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:05:54,712][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:05:55,213][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:05:55,713][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:05:56,213][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:05:56,713][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:05:57,214][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:05:57,715][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:05:58,215][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:05:58,714][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:05:59,215][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:05:59,714][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:06:00,213][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:06:00,713][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:06:01,214][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:06:01,714][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:06:02,215][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:06:02,713][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:06:03,211][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:06:03,709][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:06:04,207][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:06:04,707][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:06:05,205][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:06:05,704][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:06:06,202][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:06:06,702][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:06:07,200][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:06:07,699][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:06:08,198][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:06:08,697][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:06:09,197][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:06:09,695][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:06:10,194][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:06:10,693][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:06:11,194][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:06:11,695][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:06:12,195][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:06:12,696][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:06:13,197][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:06:13,696][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:06:14,196][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:06:14,696][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:06:15,201][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:06:15,703][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:06:16,205][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:06:16,707][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:06:17,208][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:06:17,709][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:06:18,214][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 05:06:18,867][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 05:06:19,574][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:06:19,577][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:06:19,578][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:06:20,327][__main__][INFO] - Iteration 403 took 1m 14s (9.33% Gen, 89.67% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 37m 22s. Estimated total time: 62h 9m 1s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 18s, 500 more iterations: 10h 21m 30s. [2026-03-26 05:06:20,329][__main__][INFO] - Starting iteration 403. [2026-03-26 05:06:20,729][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:06:20,730][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:06:21,334][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:06:27,491][__main__][INFO] - Number of regex retries in iteration 403: 1 [2026-03-26 05:06:27,492][__main__][INFO] - agents played in iteration 403 are Alice, Bob [2026-03-26 05:06:28,396][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:06:28,441][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:06:28,442][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:06:28,443][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:06:29,046][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:06:29,502][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:06:30,007][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:06:30,508][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:06:31,007][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:06:31,508][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:06:32,007][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:06:32,508][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:06:33,007][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:06:33,509][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:06:34,011][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:06:34,518][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:06:35,019][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:06:35,520][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:06:36,022][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:06:36,523][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:06:37,023][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:06:37,521][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:06:38,021][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:06:38,528][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:06:39,028][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:06:39,527][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:06:40,026][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:06:40,525][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:06:41,024][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:06:41,526][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:06:42,028][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:06:42,530][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:06:43,031][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:06:43,532][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:06:44,032][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:06:44,533][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:06:45,034][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:06:45,534][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:06:46,033][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:06:46,532][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:06:47,031][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:06:47,530][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:06:48,030][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:06:48,531][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:06:49,029][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:06:49,529][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:06:50,027][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:06:50,526][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:06:51,025][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:06:51,523][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:06:52,023][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:06:52,522][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:06:53,020][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:06:53,521][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:06:54,020][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:06:54,520][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:06:55,022][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:06:55,523][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:06:56,024][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:06:56,525][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:06:57,026][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:06:57,525][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:06:58,024][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:06:58,522][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:06:59,021][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:06:59,520][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:07:00,019][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:07:00,519][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:07:01,017][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:07:01,515][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:07:02,012][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:07:02,510][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:07:03,008][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:07:03,507][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:07:04,007][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:07:04,506][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:07:05,005][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:07:05,505][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:07:06,004][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:07:06,504][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:07:07,004][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:07:07,504][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:07:08,005][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:07:08,505][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:07:09,005][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:07:09,503][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:07:10,000][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:07:10,497][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:07:10,996][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:07:11,495][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:07:11,994][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:07:12,492][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:07:12,990][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:07:13,490][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:07:13,991][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:07:14,490][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:07:14,990][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:07:15,493][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:07:15,996][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:07:16,496][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:07:16,996][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:07:17,494][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:07:17,995][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:07:18,496][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:07:18,995][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:07:19,493][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:07:19,991][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:07:20,490][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:07:20,988][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:07:21,487][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:07:21,987][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:07:22,487][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:07:22,985][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:07:23,485][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:07:23,983][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:07:24,482][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:07:24,980][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:07:25,479][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:07:25,977][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:07:26,476][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:07:26,976][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:07:27,476][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:07:27,976][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:07:28,475][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:07:28,974][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:07:29,473][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:07:29,973][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:07:30,471][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:07:30,969][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:07:31,468][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:07:31,967][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:07:32,466][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:07:32,965][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 05:07:33,618][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.30%, ΔTime: 00:01:04 [2026-03-26 05:07:34,424][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:07:34,426][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:07:34,428][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:07:35,267][__main__][INFO] - Iteration 404 took 1m 14s (9.07% Gen, 89.80% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 34m 2s. Estimated total time: 62h 6m 55s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 13s, 500 more iterations: 10h 21m 9s. [2026-03-26 05:07:35,269][__main__][INFO] - Starting iteration 404. [2026-03-26 05:07:35,669][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:07:35,669][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:07:40,079][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:07:42,353][__main__][INFO] - Number of regex retries in iteration 404: 1 [2026-03-26 05:07:42,354][__main__][INFO] - agents played in iteration 404 are Alice, Bob [2026-03-26 05:07:43,268][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:07:43,313][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:07:43,314][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:07:43,315][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:07:43,928][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:07:44,386][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:07:44,891][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:07:45,393][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:07:45,894][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:07:46,394][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:07:46,895][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:07:47,396][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:07:47,901][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:07:48,404][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:07:48,905][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:07:49,405][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:07:49,904][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:07:50,405][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:07:50,906][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:07:51,407][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:07:51,907][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:07:52,410][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:07:52,910][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:07:53,412][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:07:53,912][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:07:54,412][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:07:54,913][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:07:55,415][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:07:55,915][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:07:56,418][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:07:56,923][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:07:57,427][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:07:57,929][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:07:58,431][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:07:58,933][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:07:59,435][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:07:59,937][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:08:00,437][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:08:00,947][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:08:01,449][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:08:01,951][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:08:02,452][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:08:02,952][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:08:03,453][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:08:03,954][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:08:04,453][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:08:04,953][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:08:05,455][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:08:05,955][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:08:06,456][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:08:06,956][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:08:07,455][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:08:07,954][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:08:08,451][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:08:08,948][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:08:09,445][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:08:09,942][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:08:10,438][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:08:10,934][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:08:11,431][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:08:11,926][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:08:12,424][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:08:12,923][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:08:13,421][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:08:13,919][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:08:14,417][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:08:14,915][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:08:15,414][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:08:15,913][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:08:16,413][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:08:16,913][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:08:17,413][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:08:17,915][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:08:18,418][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:08:18,918][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:08:19,418][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:08:19,918][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:08:20,416][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:08:20,914][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:08:21,413][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:08:21,913][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:08:22,411][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:08:22,910][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:08:23,409][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:08:23,907][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:08:24,406][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:08:24,903][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:08:25,403][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:08:25,902][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:08:26,401][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:08:26,900][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:08:27,398][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:08:27,896][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:08:28,396][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:08:28,895][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:08:29,396][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:08:29,896][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:08:30,397][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:08:30,896][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:08:31,396][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:08:31,895][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:08:32,394][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:08:32,893][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:08:33,393][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:08:33,892][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:08:34,391][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:08:34,890][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:08:35,389][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:08:35,889][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:08:36,387][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:08:36,886][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:08:37,385][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:08:37,884][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:08:38,383][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:08:38,881][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:08:39,379][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:08:39,876][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:08:40,371][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:08:40,867][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:08:41,362][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:08:41,858][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:08:42,353][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:08:42,848][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:08:43,343][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:08:43,839][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:08:44,336][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:08:44,835][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:08:45,334][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:08:45,834][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:08:46,333][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:08:46,833][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:08:47,333][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:08:47,831][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 05:08:48,491][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 05:08:49,275][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:08:49,278][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:08:49,279][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:08:50,033][__main__][INFO] - Iteration 405 took 1m 14s (8.99% Gen, 90.00% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 24m 7s. Estimated total time: 61h 58m 15s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 56s, 500 more iterations: 10h 19m 42s. [2026-03-26 05:08:50,036][__main__][INFO] - Starting iteration 405. [2026-03-26 05:08:50,435][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:08:50,436][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:08:52,584][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:08:53,094][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:08:54,475][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:08:57,170][__main__][INFO] - Number of regex retries in iteration 405: 3 [2026-03-26 05:08:57,171][__main__][INFO] - agents played in iteration 405 are Alice, Bob [2026-03-26 05:08:58,044][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:08:58,089][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:08:58,090][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:08:58,090][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:08:58,707][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:08:59,162][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:08:59,666][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:09:00,165][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:09:00,665][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:09:01,166][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:09:01,668][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:09:02,168][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:09:02,668][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:09:03,167][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:09:03,665][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:09:04,162][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:09:04,660][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:09:05,157][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:09:05,655][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:09:06,153][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:09:06,651][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:09:07,150][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:09:07,649][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:09:08,148][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:09:08,646][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:09:09,146][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:09:09,645][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:09:10,148][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:09:10,648][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:09:11,149][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:09:11,650][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:09:12,150][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:09:12,650][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:09:13,150][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:09:13,650][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:09:14,156][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:09:14,657][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:09:15,156][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:09:15,656][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:09:16,155][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:09:16,655][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:09:17,156][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:09:17,656][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:09:18,156][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:09:18,661][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:09:19,162][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:09:19,663][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:09:20,163][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:09:20,664][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:09:21,164][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:09:21,665][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:09:22,165][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:09:22,672][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:09:23,172][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:09:23,672][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:09:24,172][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:09:24,672][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:09:25,174][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:09:25,674][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:09:26,175][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:09:26,675][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:09:27,179][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:09:27,680][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:09:28,181][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:09:28,682][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:09:29,183][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:09:29,684][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:09:30,185][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:09:30,688][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:09:31,191][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:09:31,694][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:09:32,197][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:09:32,700][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:09:33,202][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:09:33,703][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:09:34,204][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:09:34,706][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:09:35,208][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:09:35,713][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:09:36,213][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:09:36,714][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:09:37,214][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:09:37,714][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:09:38,215][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:09:38,714][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:09:39,215][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:09:39,716][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:09:40,217][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:09:40,717][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:09:41,217][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:09:41,717][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:09:42,218][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:09:42,718][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:09:43,219][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:09:43,720][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:09:44,221][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:09:44,723][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:09:45,223][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:09:45,724][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:09:46,225][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:09:46,725][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:09:47,224][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:09:47,724][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:09:48,228][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:09:48,728][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:09:49,227][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:09:49,726][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:09:50,225][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:09:50,724][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:09:51,223][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:09:51,723][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:09:52,223][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:09:52,724][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:09:53,223][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:09:53,722][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:09:54,222][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:09:54,722][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:09:55,221][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:09:55,718][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:09:56,217][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:09:56,715][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:09:57,214][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:09:57,712][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:09:58,211][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:09:58,710][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:09:59,210][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:09:59,711][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:10:00,212][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:10:00,712][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:10:01,212][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:10:01,711][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:10:02,211][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:10:02,711][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 05:10:03,370][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 05:10:04,590][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:10:04,593][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:10:04,594][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:10:05,366][__main__][INFO] - Iteration 406 took 1m 14s (8.99% Gen, 89.98% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 51m 12s. Estimated total time: 62h 26m 35s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 53s, 500 more iterations: 10h 24m 25s. [2026-03-26 05:10:05,369][__main__][INFO] - Starting iteration 406. [2026-03-26 05:10:05,767][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:10:05,767][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:10:06,356][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:10:13,545][__main__][INFO] - Number of regex retries in iteration 406: 1 [2026-03-26 05:10:13,546][__main__][INFO] - agents played in iteration 406 are Alice, Bob [2026-03-26 05:10:14,431][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:10:14,475][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:10:14,476][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:10:14,477][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:10:15,085][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:10:15,542][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:10:16,047][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:10:16,546][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:10:17,045][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:10:17,546][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:10:18,047][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:10:18,549][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:10:19,048][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:10:19,548][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:10:20,047][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:10:20,547][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:10:21,048][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:10:21,548][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:10:22,047][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:10:22,548][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:10:23,048][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:10:23,547][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:10:24,044][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:10:24,543][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:10:25,043][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:10:25,541][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:10:26,039][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:10:26,537][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:10:27,035][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:10:27,540][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:10:28,040][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:10:28,541][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:10:29,041][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:10:29,543][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:10:30,043][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:10:30,543][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:10:31,043][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:10:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:10:32,041][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:10:32,541][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:10:33,040][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:10:33,542][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:10:34,043][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:10:34,542][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:10:35,042][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:10:35,542][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:10:36,042][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:10:36,543][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:10:37,043][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:10:37,542][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:10:38,047][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:10:38,550][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:10:39,051][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:10:39,552][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:10:40,052][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:10:40,553][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:10:41,054][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:10:41,554][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:10:42,062][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:10:42,564][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:10:43,065][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:10:43,565][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:10:44,066][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:10:44,568][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:10:45,069][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:10:45,569][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:10:46,070][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:10:46,576][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:10:47,077][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:10:47,579][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:10:48,081][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:10:48,586][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:10:49,087][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:10:49,589][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:10:50,091][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:10:50,596][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:10:51,100][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:10:51,602][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:10:52,103][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:10:52,603][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:10:53,104][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:10:53,607][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:10:54,109][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:10:54,608][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:10:55,115][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:10:55,618][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:10:56,118][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:10:56,618][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:10:57,118][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:10:57,618][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:10:58,119][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:10:58,619][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:10:59,119][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:10:59,625][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:11:00,130][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:11:00,633][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:11:01,136][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:11:01,639][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:11:02,142][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:11:02,644][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:11:03,147][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:11:03,654][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:11:04,155][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:11:04,655][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:11:05,156][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:11:05,657][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:11:06,157][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:11:06,658][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:11:07,157][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:11:07,656][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:11:08,158][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:11:08,659][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:11:09,159][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:11:09,659][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:11:10,159][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:11:10,658][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:11:11,157][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:11:11,657][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:11:12,156][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:11:12,657][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:11:13,156][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:11:13,656][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:11:14,155][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:11:14,655][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:11:15,154][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:11:15,654][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:11:16,153][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:11:16,653][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:11:17,152][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:11:17,651][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:11:18,152][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:11:18,653][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:11:19,153][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21728 tokens. [2026-03-26 05:11:19,791][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:11:20,560][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:11:20,565][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:11:20,567][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:11:21,289][__main__][INFO] - Iteration 407 took 1m 15s (10.30% Gen, 88.74% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 54h 19m 30s. Estimated total time: 62h 56m 9s. Time estimates for 10 more iterations: 12m 35s, 100 more iterations: 2h 5m 52s, 500 more iterations: 10h 29m 21s. [2026-03-26 05:11:21,291][__main__][INFO] - Starting iteration 407. [2026-03-26 05:11:21,692][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:11:21,693][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:11:28,790][__main__][INFO] - Number of regex retries in iteration 407: 0 [2026-03-26 05:11:28,791][__main__][INFO] - agents played in iteration 407 are Alice, Bob [2026-03-26 05:11:29,678][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:11:29,723][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:11:29,724][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:11:29,724][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:11:30,332][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:11:30,804][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:11:31,307][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:11:31,806][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:11:32,307][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:11:32,807][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:11:33,307][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:11:33,806][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:11:34,305][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:11:34,804][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:11:35,303][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:11:35,803][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:11:36,302][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:11:36,802][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:11:37,302][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:11:37,801][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:11:38,301][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:11:38,799][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:11:39,298][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:11:39,797][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:11:40,296][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:11:40,795][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:11:41,290][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:11:41,789][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:11:42,288][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:11:42,787][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:11:43,288][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:11:43,787][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:11:44,287][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:11:44,786][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:11:45,287][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:11:45,788][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:11:46,289][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:11:46,788][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:11:47,288][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:11:47,788][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:11:48,289][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:11:48,789][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:11:49,288][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:11:49,786][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:11:50,285][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:11:50,783][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:11:51,282][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:11:51,780][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:11:52,279][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:11:52,777][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:11:53,275][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:11:53,773][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:11:54,272][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:11:54,773][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:11:55,274][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:11:55,775][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:11:56,275][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:11:56,774][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:11:57,274][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:11:57,775][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:11:58,274][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:11:58,775][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:11:59,274][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:11:59,773][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:12:00,273][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:12:00,773][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:12:01,273][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:12:01,773][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:12:02,272][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:12:02,771][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:12:03,270][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:12:03,769][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:12:04,268][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:12:04,767][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:12:05,267][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:12:05,767][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:12:06,269][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:12:06,769][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:12:07,268][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:12:07,766][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:12:08,266][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:12:08,768][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:12:09,268][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:12:09,770][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:12:10,270][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:12:10,770][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:12:11,270][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:12:11,770][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:12:12,271][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:12:12,772][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:12:13,272][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:12:13,773][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:12:14,273][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:12:14,773][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:12:15,273][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:12:15,772][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:12:16,273][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:12:16,772][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:12:17,277][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:12:17,777][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:12:18,280][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:12:18,780][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:12:19,279][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:12:19,778][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:12:20,278][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:12:20,777][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:12:21,277][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:12:21,777][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:12:22,277][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:12:22,777][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:12:23,278][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:12:23,779][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:12:24,278][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:12:24,778][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:12:25,278][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:12:25,783][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:12:26,285][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:12:26,788][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:12:27,291][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:12:27,792][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:12:28,294][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:12:28,797][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:12:29,299][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:12:29,800][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:12:30,306][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:12:30,808][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:12:31,310][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:12:31,812][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:12:32,313][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:12:32,815][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:12:33,316][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:12:33,816][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:12:34,326][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21703 tokens. [2026-03-26 05:12:34,963][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 05:12:35,793][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:12:35,796][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:12:35,797][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:12:36,470][__main__][INFO] - Iteration 408 took 1m 14s (9.49% Gen, 89.61% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 53h 41m 2s. Estimated total time: 62h 18m 56s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 37s, 500 more iterations: 10h 23m 9s. [2026-03-26 05:12:36,473][__main__][INFO] - Starting iteration 408. [2026-03-26 05:12:36,871][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:12:36,872][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:12:43,495][__main__][INFO] - Number of regex retries in iteration 408: 0 [2026-03-26 05:12:43,495][__main__][INFO] - agents played in iteration 408 are Alice, Bob [2026-03-26 05:12:44,393][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:12:44,438][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:12:44,438][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:12:44,439][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:12:45,043][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:12:45,498][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:12:46,002][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:12:46,501][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:12:47,000][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:12:47,498][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:12:47,999][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:12:48,501][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:12:49,001][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:12:49,499][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:12:49,997][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:12:50,496][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:12:50,996][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:12:51,495][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:12:51,993][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:12:52,492][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:12:52,992][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:12:53,491][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:12:53,990][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:12:54,491][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:12:54,990][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:12:55,488][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:12:55,988][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:12:56,486][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:12:56,985][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:12:57,487][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:12:57,989][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:12:58,490][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:12:58,991][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:12:59,491][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:12:59,993][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:13:00,494][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:13:00,994][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:13:01,495][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:13:01,996][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:13:02,497][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:13:02,996][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:13:03,496][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:13:03,997][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:13:04,498][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:13:04,997][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:13:05,497][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:13:05,995][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:13:06,494][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:13:06,995][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:13:07,494][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:13:07,994][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:13:08,493][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:13:08,992][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:13:09,491][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:13:09,991][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:13:10,489][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:13:10,989][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:13:11,489][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:13:11,989][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:13:12,489][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:13:12,988][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:13:13,487][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:13:13,985][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:13:14,483][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:13:14,981][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:13:15,481][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:13:15,981][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:13:16,479][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:13:17,003][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:13:17,501][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:13:18,001][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:13:18,502][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:13:19,000][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:13:19,500][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:13:19,999][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:13:20,499][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:13:20,997][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:13:21,496][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:13:21,994][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:13:22,493][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:13:22,992][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:13:23,495][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:13:23,994][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:13:24,494][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:13:24,995][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:13:25,495][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:13:25,994][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:13:26,493][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:13:26,992][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:13:27,491][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:13:27,994][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:13:28,496][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:13:28,996][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:13:29,497][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:13:30,000][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:13:30,502][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:13:31,004][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:13:31,505][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:13:32,006][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:13:32,514][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:13:33,017][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:13:33,519][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:13:34,020][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:13:34,521][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:13:35,022][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:13:35,523][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:13:36,025][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:13:36,527][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:13:37,033][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:13:37,535][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:13:38,036][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:13:38,538][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:13:39,039][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:13:39,539][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:13:40,040][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:13:40,540][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:13:41,041][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:13:41,543][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:13:42,045][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:13:42,546][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:13:43,048][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:13:43,548][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:13:44,050][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:13:44,551][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:13:45,052][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:13:45,553][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:13:46,060][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:13:46,561][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:13:47,062][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:13:47,563][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:13:48,065][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:13:48,568][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:13:49,069][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21732 tokens. [2026-03-26 05:13:49,713][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 05:13:50,541][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:13:50,543][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:13:50,545][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:13:51,401][__main__][INFO] - Iteration 409 took 1m 14s (8.89% Gen, 89.96% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 27m 21s. Estimated total time: 62h 6m 31s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 13s, 500 more iterations: 10h 21m 5s. [2026-03-26 05:13:51,404][__main__][INFO] - Starting iteration 409. [2026-03-26 05:13:51,804][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:13:51,805][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:13:58,730][__main__][INFO] - Number of regex retries in iteration 409: 0 [2026-03-26 05:13:58,731][__main__][INFO] - agents played in iteration 409 are Alice, Bob [2026-03-26 05:13:59,627][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:13:59,672][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:13:59,673][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:13:59,674][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:14:00,280][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:14:00,735][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:14:01,239][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:14:01,738][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:14:02,238][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:14:02,738][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:14:03,238][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:14:03,738][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:14:04,239][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:14:04,736][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:14:05,234][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:14:05,733][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:14:06,231][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:14:06,730][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:14:07,229][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:14:07,727][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:14:08,227][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:14:08,726][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:14:09,225][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:14:09,723][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:14:10,222][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:14:10,720][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:14:11,218][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:14:11,717][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:14:12,216][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:14:12,715][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:14:13,213][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:14:13,712][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:14:14,212][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:14:14,711][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:14:15,210][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:14:15,710][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:14:16,209][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:14:16,710][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:14:17,211][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:14:17,712][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:14:18,215][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:14:18,716][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:14:19,216][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:14:19,716][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:14:20,217][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:14:20,716][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:14:21,215][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:14:21,715][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:14:22,214][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:14:22,713][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:14:23,212][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:14:23,711][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:14:24,209][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:14:24,708][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:14:25,208][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:14:25,706][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:14:26,206][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:14:26,704][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:14:27,202][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:14:27,701][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:14:28,203][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:14:28,701][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:14:29,201][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:14:29,700][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:14:30,199][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:14:30,698][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:14:31,197][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:14:31,696][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:14:32,196][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:14:32,696][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:14:33,196][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:14:33,696][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:14:34,196][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:14:34,696][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:14:35,195][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:14:35,696][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:14:36,197][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:14:36,696][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:14:37,195][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:14:37,693][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:14:38,193][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:14:38,692][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:14:39,192][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:14:39,690][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:14:40,190][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:14:40,690][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:14:41,189][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:14:41,689][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:14:42,188][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:14:42,686][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:14:43,185][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:14:43,683][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:14:44,181][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:14:44,679][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:14:45,179][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:14:45,681][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:14:46,181][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:14:46,679][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:14:47,177][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:14:47,676][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:14:48,178][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:14:48,679][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:14:49,179][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:14:49,678][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:14:50,178][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:14:50,678][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:14:51,177][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:14:51,678][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:14:52,178][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:14:52,677][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:14:53,175][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:14:53,674][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:14:54,174][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:14:54,672][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:14:55,171][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:14:55,670][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:14:56,169][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:14:56,666][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:14:57,164][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:14:57,662][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:14:58,160][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:14:58,657][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:14:59,155][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:14:59,653][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:15:00,151][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:15:00,650][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:15:01,149][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:15:01,648][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:15:02,146][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:15:02,645][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:15:03,143][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:15:03,642][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:15:04,142][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21706 tokens. [2026-03-26 05:15:04,778][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:15:05,567][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:15:05,569][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:15:05,571][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:15:06,306][__main__][INFO] - Iteration 410 took 1m 14s (9.30% Gen, 89.71% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 24m 46s. Estimated total time: 62h 5m 11s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 51s. [2026-03-26 05:15:06,308][__main__][INFO] - Starting iteration 410. [2026-03-26 05:15:06,710][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:15:06,710][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:15:13,665][__main__][INFO] - Number of regex retries in iteration 410: 0 [2026-03-26 05:15:13,666][__main__][INFO] - agents played in iteration 410 are Alice, Bob [2026-03-26 05:15:14,559][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:15:14,604][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:15:14,604][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:15:14,605][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:15:15,212][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:15:15,667][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:15:16,170][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:15:16,670][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:15:17,170][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:15:17,670][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:15:18,170][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:15:18,671][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:15:19,170][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:15:19,671][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:15:20,170][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:15:20,669][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:15:21,167][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:15:21,666][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:15:22,164][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:15:22,661][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:15:23,160][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:15:23,660][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:15:24,159][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:15:24,658][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:15:25,157][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:15:25,656][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:15:26,155][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:15:26,654][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:15:27,153][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:15:27,652][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:15:28,151][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:15:28,649][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:15:29,148][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:15:29,646][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:15:30,145][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:15:30,643][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:15:31,142][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:15:31,640][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:15:32,138][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:15:32,636][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:15:33,135][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:15:33,634][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:15:34,133][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:15:34,631][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:15:35,130][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:15:35,630][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:15:36,130][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:15:36,629][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:15:37,128][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:15:37,628][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:15:38,127][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:15:38,625][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:15:39,124][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:15:39,623][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:15:40,122][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:15:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:15:41,121][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:15:41,621][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:15:42,119][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:15:42,619][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:15:43,119][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:15:43,618][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:15:44,116][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:15:44,616][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:15:45,116][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:15:45,615][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:15:46,114][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:15:46,612][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:15:47,111][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:15:47,611][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:15:48,111][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:15:48,614][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:15:49,113][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:15:49,614][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:15:50,112][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:15:50,613][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:15:51,112][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:15:51,612][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:15:52,110][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:15:52,609][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:15:53,107][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:15:53,606][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:15:54,105][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:15:54,604][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:15:55,103][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:15:55,604][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:15:56,104][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:15:56,604][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:15:57,103][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:15:57,603][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:15:58,103][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:15:58,602][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:15:59,102][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:15:59,601][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:16:00,101][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:16:00,599][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:16:01,098][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:16:01,598][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:16:02,097][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:16:02,595][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:16:03,095][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:16:03,594][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:16:04,094][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:16:04,594][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:16:05,094][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:16:05,595][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:16:06,095][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:16:06,595][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:16:07,094][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:16:07,594][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:16:08,094][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:16:08,593][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:16:09,094][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:16:09,593][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:16:10,092][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:16:10,592][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:16:11,091][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:16:11,591][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:16:12,090][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:16:12,590][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:16:13,089][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:16:13,589][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:16:14,088][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:16:14,586][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:16:15,084][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:16:15,586][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:16:16,086][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:16:16,585][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:16:17,083][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:16:17,582][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:16:18,081][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:16:18,581][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:16:19,079][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 05:16:19,719][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 05:16:20,496][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:16:20,499][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:16:20,500][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:16:21,217][__main__][INFO] - Iteration 411 took 1m 14s (9.33% Gen, 89.70% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 23m 44s. Estimated total time: 62h 5m 24s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 54s. [2026-03-26 05:16:21,222][__main__][INFO] - Starting iteration 411. [2026-03-26 05:16:21,621][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:16:21,621][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:16:28,956][__main__][INFO] - Number of regex retries in iteration 411: 0 [2026-03-26 05:16:28,957][__main__][INFO] - agents played in iteration 411 are Alice, Bob [2026-03-26 05:16:29,851][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:16:29,896][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:16:29,897][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:16:29,897][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:16:30,505][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:16:30,959][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:16:31,463][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:16:31,962][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:16:32,461][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:16:32,959][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:16:33,457][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:16:33,956][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:16:34,456][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:16:34,954][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:16:35,453][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:16:35,951][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:16:36,449][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:16:36,950][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:16:37,449][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:16:37,948][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:16:38,448][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:16:38,947][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:16:39,446][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:16:39,944][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:16:40,444][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:16:40,942][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:16:41,441][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:16:41,941][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:16:42,440][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:16:42,940][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:16:43,440][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:16:43,940][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:16:44,440][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:16:44,939][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:16:45,439][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:16:45,938][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:16:46,438][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:16:46,936][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:16:47,434][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:16:47,933][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:16:48,433][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:16:48,932][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:16:49,431][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:16:49,929][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:16:50,427][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:16:50,925][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:16:51,423][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:16:51,922][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:16:52,421][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:16:52,919][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:16:53,418][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:16:53,917][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:16:54,416][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:16:54,916][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:16:55,416][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:16:55,916][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:16:56,416][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:16:56,916][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:16:57,416][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:16:57,916][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:16:58,416][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:16:58,915][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:16:59,414][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:16:59,913][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:17:00,412][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:17:00,912][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:17:01,411][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:17:01,910][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:17:02,408][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:17:02,907][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:17:03,407][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:17:03,907][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:17:04,406][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:17:04,905][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:17:05,404][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:17:05,903][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:17:06,403][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:17:06,902][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:17:07,401][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:17:07,900][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:17:08,399][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:17:08,898][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:17:09,396][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:17:09,894][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:17:10,394][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:17:10,893][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:17:11,393][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:17:11,892][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:17:12,392][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:17:12,890][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:17:13,390][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:17:13,889][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:17:14,389][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:17:14,890][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:17:15,391][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:17:15,890][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:17:16,391][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:17:16,890][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:17:17,389][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:17:17,890][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:17:18,391][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:17:18,892][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:17:19,391][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:17:19,889][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:17:20,387][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:17:20,885][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:17:21,382][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:17:21,880][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:17:22,378][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:17:22,877][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:17:23,376][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:17:23,876][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:17:24,374][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:17:24,873][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:17:25,373][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:17:25,867][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:17:26,367][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:17:26,866][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:17:27,366][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:17:27,866][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:17:28,365][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:17:28,864][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:17:29,362][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:17:29,860][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:17:30,360][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:17:30,858][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:17:31,358][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:17:31,856][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:17:32,355][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:17:32,854][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:17:33,353][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:17:33,852][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:17:34,351][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21710 tokens. [2026-03-26 05:17:34,982][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 05:17:35,764][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:17:35,767][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:17:35,769][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:17:36,424][__main__][INFO] - Iteration 412 took 1m 14s (9.81% Gen, 89.32% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 37m 17s. Estimated total time: 62h 20m 11s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 40s, 500 more iterations: 10h 23m 21s. [2026-03-26 05:17:36,426][__main__][INFO] - Starting iteration 412. [2026-03-26 05:17:36,826][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:17:36,827][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:17:37,450][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:17:43,889][__main__][INFO] - Number of regex retries in iteration 412: 1 [2026-03-26 05:17:43,889][__main__][INFO] - agents played in iteration 412 are Alice, Bob [2026-03-26 05:17:44,784][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:17:44,828][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:17:44,829][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:17:44,830][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:17:45,438][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:17:45,895][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:17:46,400][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:17:46,903][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:17:47,404][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:17:47,905][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:17:48,407][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:17:48,907][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:17:49,408][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:17:49,908][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:17:50,407][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:17:50,905][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:17:51,404][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:17:51,902][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:17:52,400][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:17:52,899][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:17:53,399][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:17:53,899][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:17:54,399][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:17:54,898][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:17:55,398][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:17:55,898][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:17:56,398][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:17:56,898][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:17:57,399][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:17:57,899][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:17:58,399][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:17:58,898][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:17:59,398][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:17:59,897][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:18:00,397][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:18:00,896][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:18:01,396][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:18:01,897][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:18:02,397][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:18:02,898][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:18:03,399][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:18:03,900][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:18:04,400][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:18:04,900][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:18:05,402][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:18:05,900][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:18:06,398][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:18:06,897][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:18:07,397][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:18:07,897][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:18:08,397][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:18:08,896][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:18:09,396][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:18:09,895][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:18:10,393][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:18:10,893][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:18:11,393][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:18:11,891][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:18:12,391][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:18:12,890][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:18:13,389][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:18:13,888][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:18:14,387][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:18:14,888][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:18:15,388][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:18:15,887][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:18:16,387][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:18:16,886][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:18:17,385][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:18:17,885][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:18:18,388][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:18:18,890][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:18:19,391][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:18:19,891][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:18:20,392][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:18:20,893][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:18:21,393][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:18:21,892][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:18:22,391][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:18:22,890][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:18:23,388][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:18:23,887][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:18:24,386][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:18:24,884][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:18:25,383][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:18:25,881][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:18:26,381][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:18:26,880][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:18:27,379][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:18:27,879][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:18:28,378][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:18:28,878][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:18:29,378][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:18:29,877][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:18:30,376][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:18:30,875][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:18:31,374][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:18:31,873][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:18:32,372][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:18:32,873][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:18:33,371][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:18:33,871][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:18:34,371][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:18:34,871][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:18:35,371][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:18:35,871][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:18:36,374][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:18:36,875][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:18:37,376][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:18:37,874][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:18:38,374][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:18:38,873][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:18:39,372][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:18:39,871][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:18:40,371][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:18:40,870][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:18:41,369][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:18:41,867][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:18:42,365][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:18:42,862][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:18:43,360][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:18:43,859][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:18:44,358][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:18:44,857][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:18:45,355][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:18:45,854][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:18:46,353][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:18:46,853][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:18:47,352][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:18:47,851][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:18:48,353][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:18:48,852][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:18:49,351][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 05:18:49,982][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 05:18:50,752][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:18:50,755][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:18:50,756][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:18:51,419][__main__][INFO] - Iteration 413 took 1m 14s (9.47% Gen, 89.64% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 25m 31s. Estimated total time: 62h 9m 40s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 19s, 500 more iterations: 10h 21m 36s. [2026-03-26 05:18:51,421][__main__][INFO] - Starting iteration 413. [2026-03-26 05:18:51,819][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:18:51,820][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:18:58,899][__main__][INFO] - Number of regex retries in iteration 413: 0 [2026-03-26 05:18:58,900][__main__][INFO] - agents played in iteration 413 are Alice, Bob [2026-03-26 05:18:59,803][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:18:59,848][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:18:59,849][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:18:59,849][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:19:00,464][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:19:00,918][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:19:01,422][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:19:01,921][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:19:02,421][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:19:02,921][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:19:03,420][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:19:03,919][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:19:04,418][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:19:04,917][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:19:05,416][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:19:05,915][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:19:06,414][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:19:06,914][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:19:07,413][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:19:07,911][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:19:08,410][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:19:08,911][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:19:09,413][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:19:09,912][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:19:10,412][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:19:10,912][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:19:11,410][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:19:11,910][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:19:12,409][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:19:12,909][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:19:13,408][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:19:13,907][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:19:14,406][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:19:14,904][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:19:15,405][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:19:15,907][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:19:16,407][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:19:16,906][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:19:17,404][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:19:17,904][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:19:18,405][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:19:18,903][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:19:19,402][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:19:19,900][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:19:20,399][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:19:20,898][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:19:21,397][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:19:21,896][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:19:22,396][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:19:22,895][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:19:23,394][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:19:23,893][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:19:24,393][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:19:24,896][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:19:25,397][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:19:25,897][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:19:26,400][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:19:26,901][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:19:27,401][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:19:27,901][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:19:28,402][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:19:28,900][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:19:29,399][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:19:29,898][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:19:30,397][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:19:30,897][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:19:31,397][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:19:31,896][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:19:32,395][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:19:32,893][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:19:33,392][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:19:33,891][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:19:34,391][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:19:34,890][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:19:35,390][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:19:35,890][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:19:36,389][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:19:36,887][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:19:37,385][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:19:37,883][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:19:38,381][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:19:38,880][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:19:39,378][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:19:39,878][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:19:40,377][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:19:40,876][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:19:41,376][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:19:41,875][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:19:42,375][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:19:42,874][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:19:43,373][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:19:43,873][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:19:44,373][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:19:44,872][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:19:45,370][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:19:45,867][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:19:46,364][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:19:46,862][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:19:47,360][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:19:47,858][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:19:48,358][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:19:48,856][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:19:49,354][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:19:49,852][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:19:50,350][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:19:50,848][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:19:51,345][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:19:51,844][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:19:52,342][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:19:52,840][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:19:53,337][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:19:53,836][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:19:54,335][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:19:54,834][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:19:55,334][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:19:55,832][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:19:56,331][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:19:56,831][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:19:57,331][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:19:57,833][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:19:58,333][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:19:58,834][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:19:59,335][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:19:59,835][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:20:00,335][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:20:00,833][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:20:01,332][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:20:01,831][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:20:02,332][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:20:02,831][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:20:03,330][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:20:03,829][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:20:04,327][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21736 tokens. [2026-03-26 05:20:04,964][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:20:05,662][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:20:05,664][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:20:05,666][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:20:06,366][__main__][INFO] - Iteration 414 took 1m 14s (9.50% Gen, 89.56% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 21m 58s. Estimated total time: 62h 7m 22s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 14s, 500 more iterations: 10h 21m 13s. [2026-03-26 05:20:06,368][__main__][INFO] - Starting iteration 414. [2026-03-26 05:20:06,769][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:20:06,770][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:20:09,196][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:20:13,793][__main__][INFO] - Number of regex retries in iteration 414: 1 [2026-03-26 05:20:13,794][__main__][INFO] - agents played in iteration 414 are Alice, Bob [2026-03-26 05:20:14,670][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:20:14,714][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:20:14,715][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:20:14,715][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:20:15,317][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:20:15,775][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:20:16,279][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:20:16,778][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:20:17,277][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:20:18,023][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:20:18,558][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:20:19,058][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:20:19,557][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:20:20,059][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:20:20,559][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:20:21,058][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:20:21,558][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:20:22,057][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:20:22,557][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:20:23,058][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:20:23,558][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:20:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:20:24,556][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:20:25,056][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:20:25,556][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:20:26,208][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:20:26,725][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:20:27,223][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:20:27,721][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:20:28,221][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:20:28,719][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:20:29,218][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:20:29,717][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:20:30,217][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:20:30,717][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:20:31,216][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:20:31,716][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:20:32,217][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:20:32,718][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:20:33,219][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:20:33,719][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:20:34,220][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:20:34,720][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:20:35,220][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:20:35,720][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:20:36,219][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:20:36,718][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:20:37,217][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:20:37,716][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:20:38,215][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:20:38,714][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:20:39,212][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:20:39,713][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:20:40,211][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:20:40,710][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:20:41,209][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:20:41,707][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:20:42,205][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:20:42,704][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:20:43,203][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:20:43,703][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:20:44,202][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:20:44,702][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:20:45,202][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:20:45,703][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:20:46,203][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:20:46,702][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:20:47,202][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:20:47,701][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:20:48,201][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:20:48,702][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:20:49,201][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:20:49,700][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:20:50,199][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:20:50,698][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:20:51,197][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:20:51,696][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:20:52,195][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:20:52,695][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:20:53,195][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:20:53,696][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:20:54,194][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:20:54,695][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:20:55,195][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:20:55,695][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:20:56,194][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:20:56,693][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:20:57,193][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:20:57,692][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:20:58,192][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:20:58,692][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:20:59,191][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:20:59,691][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:21:00,189][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:21:00,689][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:21:01,188][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:21:01,686][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:21:02,185][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:21:02,684][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:21:03,183][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:21:03,681][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:21:04,181][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:21:04,680][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:21:05,180][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:21:05,680][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:21:06,181][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:21:06,682][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:21:07,181][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:21:07,681][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:21:08,179][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:21:08,678][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:21:09,177][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:21:09,676][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:21:10,174][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:21:10,672][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:21:11,172][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:21:11,671][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:21:12,169][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:21:12,667][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:21:13,165][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:21:13,663][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:21:14,160][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:21:14,659][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:21:15,157][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:21:15,655][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:21:16,154][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:21:16,652][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:21:17,150][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:21:17,650][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:21:18,150][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:21:18,651][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:21:19,151][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:21:19,649][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 05:21:20,635][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:05 [2026-03-26 05:21:21,377][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:21:21,380][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:21:21,381][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:21:22,071][__main__][INFO] - Iteration 415 took 1m 15s (9.33% Gen, 89.75% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 53h 58m 26s. Estimated total time: 62h 45m 6s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 30s, 500 more iterations: 10h 27m 31s. [2026-03-26 05:21:22,074][__main__][INFO] - Starting iteration 415. [2026-03-26 05:21:22,475][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:21:22,475][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:21:27,753][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:21:29,906][__main__][INFO] - Number of regex retries in iteration 415: 1 [2026-03-26 05:21:29,907][__main__][INFO] - agents played in iteration 415 are Alice, Bob [2026-03-26 05:21:31,048][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:21:31,093][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:21:31,094][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:21:31,094][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:21:31,701][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:21:32,160][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:21:32,664][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:21:33,164][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:21:33,662][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:21:34,161][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:21:34,660][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:21:35,159][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:21:35,658][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:21:36,158][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:21:36,656][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:21:37,154][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:21:37,652][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:21:38,151][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:21:38,650][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:21:39,148][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:21:39,646][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:21:40,145][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:21:40,644][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:21:41,143][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:21:41,642][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:21:42,141][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:21:42,640][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:21:43,139][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:21:43,638][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:21:44,137][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:21:44,635][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:21:45,133][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:21:45,632][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:21:46,131][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:21:46,629][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:21:47,129][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:21:47,627][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:21:48,127][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:21:48,628][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:21:49,127][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:21:49,627][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:21:50,126][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:21:50,625][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:21:51,123][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:21:51,624][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:21:52,122][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:21:52,619][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:21:53,118][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:21:53,618][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:21:54,117][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:21:54,615][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:21:55,114][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:21:55,613][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:21:56,112][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:21:56,612][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:21:57,110][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:21:57,610][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:21:58,110][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:21:58,609][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:21:59,107][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:21:59,607][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:22:00,106][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:22:00,605][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:22:01,103][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:22:01,602][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:22:02,102][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:22:02,602][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:22:03,101][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:22:03,600][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:22:04,099][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:22:04,598][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:22:05,097][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:22:05,597][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:22:06,095][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:22:06,595][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:22:07,094][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:22:07,594][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:22:08,093][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:22:08,594][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:22:09,094][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:22:09,594][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:22:10,093][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:22:10,592][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:22:11,092][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:22:11,592][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:22:12,092][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:22:12,594][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:22:13,094][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:22:13,594][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:22:14,095][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:22:14,594][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:22:15,094][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:22:15,599][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:22:16,100][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:22:16,600][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:22:17,098][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:22:17,598][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:22:18,097][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:22:18,598][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:22:19,097][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:22:19,596][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:22:20,096][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:22:20,596][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:22:21,095][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:22:21,595][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:22:22,095][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:22:22,595][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:22:23,094][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:22:23,594][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:22:24,094][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:22:24,593][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:22:25,092][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:22:25,590][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:22:26,088][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:22:26,585][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:22:27,083][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:22:27,581][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:22:28,081][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:22:28,580][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:22:29,079][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:22:29,578][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:22:30,078][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:22:30,577][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:22:31,076][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:22:31,576][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:22:32,076][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:22:32,575][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:22:33,073][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:22:33,572][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:22:34,071][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:22:34,570][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:22:35,070][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:22:35,569][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 05:22:36,752][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:05 [2026-03-26 05:22:37,551][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:22:37,553][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:22:37,555][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:22:38,349][__main__][INFO] - Iteration 416 took 1m 15s (9.79% Gen, 89.16% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 54h 25m 47s. Estimated total time: 63h 13m 44s. Time estimates for 10 more iterations: 12m 38s, 100 more iterations: 2h 6m 27s, 500 more iterations: 10h 32m 17s. [2026-03-26 05:22:38,351][__main__][INFO] - Starting iteration 416. [2026-03-26 05:22:38,750][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:22:38,751][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:22:41,808][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:22:44,583][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:22:46,205][__main__][INFO] - Number of regex retries in iteration 416: 2 [2026-03-26 05:22:46,206][__main__][INFO] - agents played in iteration 416 are Alice, Bob [2026-03-26 05:22:47,101][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:22:47,146][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:22:47,146][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:22:47,147][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:22:47,754][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:22:48,212][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:22:48,716][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:22:49,215][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:22:49,714][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:22:50,213][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:22:50,712][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:22:51,210][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:22:51,709][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:22:52,209][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:22:52,707][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:22:53,207][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:22:53,706][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:22:54,205][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:22:54,704][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:22:55,204][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:22:55,703][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:22:56,202][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:22:56,701][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:22:57,200][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:22:57,700][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:22:58,200][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:22:58,699][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:22:59,198][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:22:59,698][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:23:00,200][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:23:00,701][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:23:01,202][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:23:01,701][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:23:02,201][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:23:02,700][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:23:03,200][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:23:03,701][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:23:04,200][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:23:04,699][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:23:05,198][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:23:05,697][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:23:06,196][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:23:06,696][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:23:07,195][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:23:07,694][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:23:08,194][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:23:08,693][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:23:09,193][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:23:09,693][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:23:10,194][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:23:10,694][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:23:11,194][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:23:11,694][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:23:12,212][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:23:12,716][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:23:13,217][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:23:13,719][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:23:14,220][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:23:14,720][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:23:15,221][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:23:15,721][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:23:16,221][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:23:16,720][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:23:17,220][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:23:17,719][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:23:18,219][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:23:18,721][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:23:19,220][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:23:19,719][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:23:20,218][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:23:20,716][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:23:21,215][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:23:21,714][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:23:22,213][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:23:22,712][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:23:23,210][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:23:23,709][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:23:24,207][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:23:24,706][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:23:25,206][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:23:25,704][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:23:26,203][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:23:26,701][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:23:27,199][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:23:27,698][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:23:28,197][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:23:28,698][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:23:29,197][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:23:29,697][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:23:30,196][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:23:30,695][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:23:31,195][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:23:31,695][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:23:32,195][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:23:32,695][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:23:33,195][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:23:33,695][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:23:34,195][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:23:34,695][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:23:35,195][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:23:35,695][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:23:36,194][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:23:36,693][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:23:37,194][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:23:37,693][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:23:38,193][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:23:38,693][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:23:39,193][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:23:39,693][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:23:40,192][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:23:40,692][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:23:41,190][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:23:41,688][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:23:42,186][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:23:42,685][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:23:43,184][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:23:43,683][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:23:44,184][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:23:44,684][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:23:45,185][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:23:45,688][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:23:46,191][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:23:46,695][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:23:47,196][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:23:47,697][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:23:48,199][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:23:48,700][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:23:49,199][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:23:49,699][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:23:50,199][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:23:50,699][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:23:51,198][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:23:51,698][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21741 tokens. [2026-03-26 05:23:52,336][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 05:23:53,469][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:23:53,472][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:23:53,473][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:23:54,140][__main__][INFO] - Iteration 417 took 1m 15s (9.89% Gen, 89.23% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 54h 0m 19s. Estimated total time: 62h 49m 31s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 39s, 500 more iterations: 10h 28m 15s. [2026-03-26 05:23:54,142][__main__][INFO] - Starting iteration 417. [2026-03-26 05:23:54,541][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:23:54,542][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:24:01,002][__main__][INFO] - Number of regex retries in iteration 417: 0 [2026-03-26 05:24:01,003][__main__][INFO] - agents played in iteration 417 are Alice, Bob [2026-03-26 05:24:01,918][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:24:02,247][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:24:02,248][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:24:02,248][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:24:02,852][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:24:03,309][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:24:03,813][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:24:04,313][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:24:04,814][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:24:05,315][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:24:05,815][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:24:06,315][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:24:06,814][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:24:07,314][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:24:07,814][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:24:08,315][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:24:08,816][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:24:09,316][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:24:09,816][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:24:10,316][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:24:10,816][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:24:11,318][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:24:11,819][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:24:12,320][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:24:12,821][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:24:13,322][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:24:13,822][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:24:14,322][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:24:14,824][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:24:15,323][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:24:15,823][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:24:16,323][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:24:16,823][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:24:17,321][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:24:17,822][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:24:18,324][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:24:18,824][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:24:19,323][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:24:19,822][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:24:20,321][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:24:20,820][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:24:21,319][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:24:21,819][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:24:22,319][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:24:22,819][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:24:23,320][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:24:23,821][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:24:24,322][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:24:24,822][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:24:25,322][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:24:25,822][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:24:26,323][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:24:26,822][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:24:27,322][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:24:27,821][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:24:28,321][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:24:28,820][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:24:29,320][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:24:29,819][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:24:30,319][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:24:30,819][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:24:31,318][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:24:31,817][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:24:32,316][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:24:32,815][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:24:33,313][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:24:33,812][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:24:34,311][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:24:34,809][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:24:35,309][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:24:35,808][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:24:36,307][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:24:36,808][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:24:37,307][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:24:37,808][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:24:38,307][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:24:38,807][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:24:39,307][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:24:39,806][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:24:40,306][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:24:40,806][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:24:41,306][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:24:41,806][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:24:42,306][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:24:42,805][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:24:43,306][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:24:43,806][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:24:44,306][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:24:44,807][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:24:45,309][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:24:45,809][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:24:46,309][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:24:46,810][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:24:47,310][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:24:47,810][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:24:48,312][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:24:48,812][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:24:49,312][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:24:49,811][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:24:50,311][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:24:50,811][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:24:51,312][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:24:51,812][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:24:52,310][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:24:52,809][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:24:53,308][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:24:53,807][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:24:54,307][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:24:54,805][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:24:55,307][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:24:55,808][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:24:56,309][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:24:56,809][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:24:57,310][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:24:57,812][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:24:58,313][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:24:58,815][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:24:59,314][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:24:59,814][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:25:00,314][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:25:00,816][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:25:01,316][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:25:01,817][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:25:02,317][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:25:02,816][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:25:03,315][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:25:03,813][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:25:04,312][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:25:04,811][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:25:05,309][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:25:05,808][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:25:06,306][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:25:06,805][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 05:25:07,433][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 05:25:08,260][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:25:08,262][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:25:08,264][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:25:09,029][__main__][INFO] - Iteration 418 took 1m 14s (8.67% Gen, 90.30% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 13m 59s. Estimated total time: 62h 4m 25s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 8s, 500 more iterations: 10h 20m 44s. [2026-03-26 05:25:09,031][__main__][INFO] - Starting iteration 418. [2026-03-26 05:25:09,430][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:25:09,431][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:25:17,866][__main__][INFO] - Number of regex retries in iteration 418: 0 [2026-03-26 05:25:17,867][__main__][INFO] - agents played in iteration 418 are Alice, Bob [2026-03-26 05:25:18,752][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:25:18,797][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:25:18,798][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:25:18,798][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:25:19,397][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:25:19,852][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:25:20,354][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:25:20,852][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:25:21,352][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:25:21,849][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:25:22,346][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:25:22,843][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:25:23,341][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:25:23,842][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:25:24,342][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:25:24,841][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:25:25,341][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:25:25,840][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:25:26,339][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:25:26,838][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:25:27,337][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:25:27,836][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:25:28,335][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:25:28,834][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:25:29,333][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:25:29,831][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:25:30,329][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:25:30,827][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:25:31,325][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:25:31,833][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:25:32,334][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:25:32,834][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:25:33,335][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:25:33,835][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:25:34,336][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:25:34,836][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:25:35,336][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:25:35,835][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:25:36,334][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:25:36,833][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:25:37,332][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:25:37,830][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:25:38,328][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:25:38,828][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:25:39,326][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:25:39,825][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:25:40,323][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:25:40,822][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:25:41,320][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:25:41,820][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:25:42,319][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:25:42,819][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:25:43,318][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:25:43,817][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:25:44,315][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:25:44,814][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:25:45,311][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:25:45,809][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:25:46,308][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:25:46,806][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:25:47,305][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:25:47,804][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:25:48,305][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:25:48,804][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:25:49,302][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:25:49,800][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:25:50,299][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:25:50,798][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:25:51,297][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:25:51,795][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:25:52,292][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:25:52,790][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:25:53,287][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:25:53,785][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:25:54,282][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:25:54,780][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:25:55,281][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:25:55,780][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:25:56,279][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:25:56,778][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:25:57,277][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:25:57,778][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:25:58,277][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:25:58,777][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:25:59,276][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:25:59,775][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:26:00,273][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:26:00,771][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:26:01,268][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:26:01,766][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:26:02,264][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:26:02,762][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:26:03,260][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:26:03,760][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:26:04,260][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:26:04,760][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:26:05,260][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:26:05,759][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:26:06,260][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:26:06,760][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:26:07,260][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:26:07,759][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:26:08,258][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:26:08,758][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:26:09,257][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:26:09,756][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:26:10,256][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:26:10,756][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:26:11,255][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:26:11,753][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:26:12,252][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:26:12,750][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:26:13,249][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:26:13,747][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:26:14,246][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:26:14,744][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:26:15,242][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:26:15,741][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:26:16,239][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:26:16,736][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:26:17,234][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:26:17,733][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:26:18,232][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:26:18,732][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:26:19,230][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:26:19,728][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:26:20,226][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:26:20,725][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:26:21,223][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:26:21,721][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:26:22,220][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:26:22,719][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:26:23,218][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 05:26:23,844][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 05:26:24,566][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:26:24,569][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:26:24,570][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:26:25,245][__main__][INFO] - Iteration 419 took 1m 15s (11.13% Gen, 87.98% Train). Generation: 8s, Training: 1m 6s. Estimated remaining time: 54h 19m 2s. Estimated total time: 63h 10m 45s. Time estimates for 10 more iterations: 12m 38s, 100 more iterations: 2h 6m 21s, 500 more iterations: 10h 31m 47s. [2026-03-26 05:26:25,247][__main__][INFO] - Starting iteration 419. [2026-03-26 05:26:25,656][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:26:25,657][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:26:32,766][__main__][INFO] - Number of regex retries in iteration 419: 0 [2026-03-26 05:26:32,767][__main__][INFO] - agents played in iteration 419 are Alice, Bob [2026-03-26 05:26:33,661][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:26:33,706][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:26:33,707][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:26:33,707][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:26:34,311][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:26:34,768][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:26:35,272][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:26:35,772][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:26:36,273][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:26:36,773][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:26:37,273][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:26:37,774][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:26:38,275][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:26:38,774][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:26:39,272][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:26:39,771][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:26:40,269][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:26:40,768][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:26:41,268][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:26:41,766][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:26:42,266][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:26:42,764][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:26:43,262][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:26:43,761][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:26:44,259][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:26:44,758][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:26:45,258][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:26:45,759][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:26:46,260][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:26:46,760][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:26:47,260][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:26:47,758][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:26:48,260][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:26:48,761][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:26:49,260][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:26:49,758][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:26:50,257][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:26:50,761][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:26:51,261][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:26:51,764][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:26:52,265][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:26:52,766][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:26:53,268][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:26:53,770][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:26:54,272][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:26:54,772][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:26:55,274][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:26:55,773][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:26:56,272][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:26:56,771][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:26:57,269][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:26:57,767][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:26:58,265][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:26:58,764][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:26:59,263][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:26:59,762][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:27:00,261][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:27:00,760][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:27:01,259][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:27:01,759][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:27:02,259][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:27:02,758][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:27:03,257][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:27:03,757][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:27:04,256][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:27:04,755][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:27:05,254][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:27:05,753][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:27:06,254][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:27:06,755][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:27:07,256][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:27:07,755][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:27:08,256][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:27:08,756][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:27:09,255][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:27:09,757][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:27:10,257][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:27:10,756][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:27:11,253][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:27:11,752][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:27:12,251][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:27:12,749][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:27:13,248][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:27:13,746][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:27:14,246][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:27:14,744][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:27:15,243][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:27:15,741][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:27:16,240][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:27:16,738][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:27:17,237][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:27:17,736][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:27:18,237][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:27:18,738][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:27:19,238][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:27:19,737][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:27:20,236][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:27:20,735][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:27:21,235][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:27:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:27:22,234][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:27:22,736][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:27:23,237][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:27:23,739][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:27:24,242][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:27:24,744][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:27:25,247][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:27:25,748][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:27:26,250][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:27:26,748][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:27:27,246][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:27:27,745][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:27:28,243][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:27:28,742][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:27:29,240][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:27:29,739][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:27:30,238][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:27:30,737][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:27:31,236][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:27:31,736][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:27:32,236][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:27:32,734][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:27:33,233][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:27:33,732][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:27:34,232][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:27:34,730][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:27:35,228][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:27:35,726][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:27:36,225][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:27:36,724][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:27:37,223][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:27:37,722][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:27:38,221][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21742 tokens. [2026-03-26 05:27:38,846][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.53%, ΔTime: 00:01:04 [2026-03-26 05:27:39,648][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:27:39,651][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:27:39,652][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:27:40,524][__main__][INFO] - Iteration 420 took 1m 14s (9.50% Gen, 89.34% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 30m 30s. Estimated total time: 62h 23m 28s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 46s, 500 more iterations: 10h 23m 54s. [2026-03-26 05:27:40,526][__main__][INFO] - Starting iteration 420. [2026-03-26 05:27:40,927][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:27:40,927][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:27:47,999][__main__][INFO] - Number of regex retries in iteration 420: 0 [2026-03-26 05:27:48,000][__main__][INFO] - agents played in iteration 420 are Alice, Bob [2026-03-26 05:27:48,901][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:27:48,946][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:27:48,947][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:27:48,947][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:27:49,547][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:27:50,002][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:27:50,506][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:27:51,008][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:27:51,507][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:27:52,007][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:27:52,506][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:27:53,006][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:27:53,507][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:27:54,007][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:27:54,508][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:27:55,009][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:27:55,509][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:27:56,008][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:27:56,508][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:27:57,008][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:27:57,509][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:27:58,009][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:27:58,507][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:27:59,006][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:27:59,505][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:28:00,003][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:28:00,502][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:28:01,001][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:28:01,500][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:28:01,999][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:28:02,498][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:28:02,996][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:28:03,496][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:28:03,995][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:28:04,495][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:28:04,994][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:28:05,493][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:28:05,990][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:28:06,491][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:28:06,989][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:28:07,488][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:28:07,986][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:28:08,486][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:28:08,985][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:28:09,485][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:28:09,983][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:28:10,481][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:28:10,978][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:28:11,477][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:28:11,975][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:28:12,472][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:28:12,969][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:28:13,465][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:28:13,967][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:28:14,466][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:28:14,967][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:28:15,469][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:28:15,971][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:28:16,471][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:28:16,970][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:28:17,470][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:28:17,966][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:28:18,463][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:28:18,958][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:28:19,453][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:28:19,950][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:28:20,445][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:28:20,940][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:28:21,436][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:28:21,935][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:28:22,435][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:28:22,934][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:28:23,434][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:28:23,933][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:28:24,431][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:28:24,930][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:28:25,430][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:28:25,929][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:28:26,427][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:28:26,926][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:28:27,425][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:28:27,923][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:28:28,423][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:28:28,922][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:28:29,421][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:28:29,920][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:28:30,418][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:28:30,918][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:28:31,416][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:28:31,915][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:28:32,413][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:28:32,912][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:28:33,410][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:28:33,908][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:28:34,407][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:28:34,905][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:28:35,404][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:28:35,902][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:28:36,400][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:28:36,899][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:28:37,398][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:28:37,898][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:28:38,398][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:28:38,898][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:28:39,398][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:28:39,898][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:28:40,398][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:28:40,898][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:28:41,398][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:28:41,896][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:28:42,396][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:28:42,895][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:28:43,395][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:28:43,895][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:28:44,394][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:28:44,894][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:28:45,393][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:28:45,892][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:28:46,391][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:28:46,890][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:28:47,389][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:28:47,888][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:28:48,389][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:28:48,890][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:28:49,390][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:28:49,885][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:28:50,380][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:28:50,875][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:28:51,371][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:28:51,866][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:28:52,363][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:28:52,858][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:28:53,353][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21584 tokens. [2026-03-26 05:28:53,979][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 05:28:54,738][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:28:54,740][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:28:54,741][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:28:55,435][__main__][INFO] - Iteration 421 took 1m 14s (9.49% Gen, 89.57% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 11m 14s. Estimated total time: 62h 5m 27s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 54s. [2026-03-26 05:28:55,437][__main__][INFO] - Starting iteration 421. [2026-03-26 05:28:55,837][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:28:55,838][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:28:56,415][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:29:02,321][__main__][INFO] - Number of regex retries in iteration 421: 1 [2026-03-26 05:29:02,322][__main__][INFO] - agents played in iteration 421 are Alice, Bob [2026-03-26 05:29:03,227][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:29:03,272][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:29:03,273][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:29:03,273][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:29:03,880][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:29:04,334][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:29:04,837][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:29:05,336][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:29:05,835][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:29:06,335][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:29:06,833][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:29:07,332][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:29:07,830][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:29:08,329][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:29:08,826][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:29:09,323][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:29:09,821][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:29:10,321][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:29:10,818][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:29:11,316][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:29:11,813][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:29:12,314][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:29:12,816][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:29:13,316][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:29:13,816][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:29:14,316][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:29:14,816][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:29:15,315][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:29:15,815][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:29:16,316][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:29:16,816][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:29:17,315][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:29:17,814][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:29:18,316][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:29:18,814][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:29:19,313][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:29:19,812][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:29:20,313][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:29:20,813][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:29:21,313][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:29:21,812][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:29:22,312][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:29:22,810][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:29:23,309][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:29:23,808][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:29:24,307][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:29:24,806][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:29:25,305][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:29:25,803][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:29:26,302][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:29:26,801][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:29:27,300][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:29:27,799][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:29:28,298][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:29:28,798][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:29:29,298][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:29:29,798][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:29:30,298][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:29:30,798][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:29:31,297][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:29:31,797][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:29:32,298][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:29:32,799][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:29:33,299][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:29:33,800][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:29:34,299][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:29:34,801][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:29:35,302][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:29:35,802][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:29:36,301][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:29:36,801][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:29:37,300][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:29:37,799][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:29:38,298][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:29:38,797][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:29:39,296][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:29:39,796][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:29:40,295][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:29:40,793][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:29:41,291][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:29:41,789][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:29:42,287][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:29:42,786][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:29:43,285][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:29:43,783][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:29:44,282][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:29:44,782][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:29:45,283][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:29:45,786][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:29:46,287][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:29:46,787][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:29:47,286][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:29:47,786][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:29:48,286][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:29:48,785][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:29:49,283][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:29:49,782][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:29:50,281][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:29:50,779][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:29:51,277][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:29:51,775][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:29:52,274][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:29:52,774][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:29:53,273][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:29:53,773][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:29:54,272][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:29:54,771][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:29:55,270][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:29:55,770][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:29:56,267][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:29:56,765][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:29:57,264][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:29:57,764][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:29:58,262][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:29:58,761][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:29:59,260][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:29:59,759][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:30:00,258][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:30:00,757][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:30:01,258][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:30:01,757][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:30:02,257][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:30:02,756][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:30:03,255][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:30:03,754][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:30:04,254][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:30:04,755][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:30:05,256][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:30:05,756][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:30:06,257][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:30:06,758][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:30:07,257][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:30:07,757][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 05:30:08,390][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:30:09,147][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:30:09,149][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:30:09,150][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:30:09,993][__main__][INFO] - Iteration 422 took 1m 14s (8.74% Gen, 90.12% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 52m 22s. Estimated total time: 61h 47m 50s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 58s. [2026-03-26 05:30:09,997][__main__][INFO] - Starting iteration 422. [2026-03-26 05:30:10,399][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:30:10,400][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:30:14,018][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:30:17,322][__main__][INFO] - Number of regex retries in iteration 422: 1 [2026-03-26 05:30:17,323][__main__][INFO] - agents played in iteration 422 are Alice, Bob [2026-03-26 05:30:18,201][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:30:18,246][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:30:18,247][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:30:18,247][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:30:18,856][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:30:19,311][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:30:19,815][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:30:20,316][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:30:20,816][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:30:21,314][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:30:21,813][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:30:22,312][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:30:22,811][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:30:23,310][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:30:23,810][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:30:24,309][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:30:24,809][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:30:25,309][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:30:25,808][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:30:26,306][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:30:26,806][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:30:27,306][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:30:27,807][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:30:28,307][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:30:28,807][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:30:29,307][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:30:29,806][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:30:30,307][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:30:30,807][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:30:31,307][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:30:31,808][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:30:32,307][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:30:32,806][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:30:33,305][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:30:33,805][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:30:34,305][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:30:34,805][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:30:35,304][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:30:35,804][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:30:36,304][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:30:36,804][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:30:37,303][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:30:37,803][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:30:38,302][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:30:38,802][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:30:39,304][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:30:39,805][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:30:40,305][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:30:40,806][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:30:41,307][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:30:41,808][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:30:42,309][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:30:42,810][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:30:43,310][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:30:43,809][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:30:44,309][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:30:44,808][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:30:45,308][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:30:45,807][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:30:46,307][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:30:46,806][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:30:47,306][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:30:47,806][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:30:48,307][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:30:48,807][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:30:49,306][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:30:49,805][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:30:50,305][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:30:50,804][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:30:51,304][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:30:51,803][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:30:52,302][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:30:52,801][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:30:53,300][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:30:53,799][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:30:54,298][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:30:54,797][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:30:55,298][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:30:55,797][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:30:56,296][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:30:56,796][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:30:57,296][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:30:57,796][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:30:58,296][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:30:58,795][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:30:59,297][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:30:59,797][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:31:00,297][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:31:00,797][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:31:01,298][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:31:01,798][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:31:02,299][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:31:02,798][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:31:03,298][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:31:03,797][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:31:04,295][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:31:04,795][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:31:05,295][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:31:05,796][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:31:06,296][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:31:06,796][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:31:07,296][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:31:07,795][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:31:08,294][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:31:08,793][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:31:09,292][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:31:09,791][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:31:10,290][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:31:10,788][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:31:11,289][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:31:11,788][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:31:12,288][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:31:12,788][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:31:13,288][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:31:13,787][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:31:14,287][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:31:14,787][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:31:15,287][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:31:15,790][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:31:16,291][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:31:16,790][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:31:17,290][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:31:17,790][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:31:18,291][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:31:18,791][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:31:19,288][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:31:19,787][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:31:20,286][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:31:20,785][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:31:21,286][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:31:21,784][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:31:22,283][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:31:22,783][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21733 tokens. [2026-03-26 05:31:23,420][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:31:24,159][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:31:24,161][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:31:24,163][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:31:24,997][__main__][INFO] - Iteration 423 took 1m 14s (9.28% Gen, 89.60% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 13m 14s. Estimated total time: 62h 9m 57s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 19s, 500 more iterations: 10h 21m 39s. [2026-03-26 05:31:25,001][__main__][INFO] - Starting iteration 423. [2026-03-26 05:31:25,401][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:31:25,402][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:31:25,996][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:31:32,180][__main__][INFO] - Number of regex retries in iteration 423: 1 [2026-03-26 05:31:32,181][__main__][INFO] - agents played in iteration 423 are Alice, Bob [2026-03-26 05:31:33,065][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:31:33,110][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:31:33,111][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:31:33,112][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:31:33,716][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:31:34,174][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:31:34,677][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:31:35,177][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:31:35,676][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:31:36,177][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:31:36,677][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:31:37,176][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:31:37,675][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:31:38,174][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:31:38,674][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:31:39,172][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:31:39,671][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:31:40,170][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:31:40,668][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:31:41,167][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:31:41,665][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:31:42,163][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:31:42,662][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:31:43,160][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:31:43,659][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:31:44,157][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:31:44,655][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:31:45,154][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:31:45,652][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:31:46,149][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:31:46,648][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:31:47,146][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:31:47,644][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:31:48,144][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:31:48,645][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:31:49,144][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:31:49,642][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:31:50,143][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:31:50,643][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:31:51,143][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:31:51,643][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:31:52,143][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:31:52,644][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:31:53,144][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:31:53,643][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:31:54,143][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:31:54,643][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:31:55,144][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:31:55,644][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:31:56,145][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:31:56,645][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:31:57,145][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:31:57,646][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:31:58,145][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:31:58,644][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:31:59,144][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:31:59,644][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:32:00,142][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:32:00,640][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:32:01,139][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:32:01,638][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:32:02,138][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:32:02,639][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:32:03,139][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:32:03,640][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:32:04,140][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:32:04,640][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:32:05,140][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:32:05,639][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:32:06,139][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:32:06,638][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:32:07,137][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:32:07,635][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:32:08,134][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:32:08,632][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:32:09,130][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:32:09,628][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:32:10,126][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:32:10,626][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:32:11,125][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:32:11,624][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:32:12,122][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:32:12,621][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:32:13,120][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:32:13,618][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:32:14,118][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:32:14,617][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:32:15,116][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:32:15,614][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:32:16,112][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:32:16,610][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:32:17,108][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:32:17,606][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:32:18,106][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:32:18,606][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:32:19,106][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:32:19,603][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:32:20,101][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:32:20,599][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:32:21,099][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:32:21,597][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:32:22,097][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:32:22,598][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:32:23,097][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:32:23,598][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:32:24,097][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:32:24,599][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:32:25,099][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:32:25,600][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:32:26,100][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:32:26,600][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:32:27,100][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:32:27,600][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:32:28,101][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:32:28,602][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:32:29,103][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:32:29,604][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:32:30,103][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:32:30,598][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:32:31,098][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:32:31,593][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:32:32,092][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:32:32,588][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:32:33,084][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:32:33,580][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:32:34,080][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:32:34,580][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:32:35,079][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:32:35,579][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:32:36,080][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:32:36,581][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:32:37,082][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:32:37,582][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21607 tokens. [2026-03-26 05:32:38,213][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 05:32:38,935][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:32:38,937][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:32:38,939][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:32:39,747][__main__][INFO] - Iteration 424 took 1m 14s (9.12% Gen, 89.79% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 59m 22s. Estimated total time: 61h 57m 20s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 54s, 500 more iterations: 10h 19m 33s. [2026-03-26 05:32:39,751][__main__][INFO] - Starting iteration 424. [2026-03-26 05:32:40,195][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:32:40,195][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:32:42,379][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:32:47,762][__main__][INFO] - Number of regex retries in iteration 424: 1 [2026-03-26 05:32:47,762][__main__][INFO] - agents played in iteration 424 are Alice, Bob [2026-03-26 05:32:48,655][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:32:48,700][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:32:48,701][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:32:48,701][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:32:49,304][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:32:49,758][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:32:50,263][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:32:50,763][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:32:51,261][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:32:51,761][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:32:52,260][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:32:52,758][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:32:53,257][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:32:53,758][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:32:54,258][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:32:54,758][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:32:55,259][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:32:55,759][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:32:56,259][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:32:56,758][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:32:57,258][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:32:57,757][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:32:58,255][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:32:58,754][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:32:59,253][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:32:59,752][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:33:00,251][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:33:00,749][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:33:01,248][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:33:01,746][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:33:02,245][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:33:02,743][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:33:03,242][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:33:03,741][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:33:04,239][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:33:04,738][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:33:05,238][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:33:05,736][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:33:06,234][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:33:06,731][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:33:07,228][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:33:07,726][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:33:08,224][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:33:08,722][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:33:09,220][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:33:09,720][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:33:10,219][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:33:10,719][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:33:11,218][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:33:11,718][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:33:12,219][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:33:12,719][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:33:13,219][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:33:13,719][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:33:14,218][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:33:14,716][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:33:15,215][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:33:15,716][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:33:16,216][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:33:16,715][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:33:17,215][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:33:17,715][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:33:18,217][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:33:18,718][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:33:19,218][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:33:19,718][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:33:20,218][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:33:20,718][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:33:21,218][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:33:21,717][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:33:22,215][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:33:22,715][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:33:23,214][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:33:23,714][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:33:24,212][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:33:24,711][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:33:25,210][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:33:25,710][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:33:26,211][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:33:26,712][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:33:27,213][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:33:27,714][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:33:28,214][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:33:28,714][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:33:29,214][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:33:29,715][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:33:30,214][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:33:30,713][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:33:31,212][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:33:31,711][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:33:32,209][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:33:32,708][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:33:33,207][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:33:33,706][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:33:34,205][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:33:34,705][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:33:35,204][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:33:35,703][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:33:36,202][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:33:36,701][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:33:37,202][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:33:37,701][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:33:38,200][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:33:38,699][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:33:39,198][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:33:39,698][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:33:40,196][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:33:40,695][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:33:41,196][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:33:41,697][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:33:42,197][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:33:42,696][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:33:43,195][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:33:43,694][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:33:44,193][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:33:44,693][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:33:45,192][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:33:45,692][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:33:46,192][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:33:46,690][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:33:47,189][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:33:47,687][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:33:48,188][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:33:48,689][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:33:49,189][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:33:49,689][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:33:50,188][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:33:50,687][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:33:51,186][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:33:51,685][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:33:52,184][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:33:52,684][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:33:53,184][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21745 tokens. [2026-03-26 05:33:53,813][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:33:54,557][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:33:54,559][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:33:54,561][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:33:55,395][__main__][INFO] - Iteration 425 took 1m 15s (10.06% Gen, 88.83% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 40m 49s. Estimated total time: 62h 40m 2s. Time estimates for 10 more iterations: 12m 32s, 100 more iterations: 2h 5m 20s, 500 more iterations: 10h 26m 40s. [2026-03-26 05:33:55,397][__main__][INFO] - Starting iteration 425. [2026-03-26 05:33:55,804][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:33:55,804][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:33:56,989][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:33:56,990][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:33:57,384][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:34:02,751][__main__][INFO] - Number of regex retries in iteration 425: 3 [2026-03-26 05:34:02,752][__main__][INFO] - agents played in iteration 425 are Alice, Bob [2026-03-26 05:34:03,652][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:34:03,697][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:34:03,697][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:34:03,698][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:34:04,320][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:34:04,774][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:34:05,277][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:34:05,777][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:34:06,276][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:34:06,775][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:34:07,272][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:34:07,770][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:34:08,267][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:34:08,768][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:34:09,267][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:34:09,765][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:34:10,265][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:34:10,763][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:34:11,262][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:34:11,761][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:34:12,259][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:34:12,757][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:34:13,258][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:34:13,757][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:34:14,256][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:34:14,755][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:34:15,254][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:34:15,755][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:34:16,254][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:34:16,752][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:34:17,251][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:34:17,748][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:34:18,248][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:34:18,746][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:34:19,244][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:34:19,742][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:34:20,240][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:34:20,741][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:34:21,240][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:34:21,739][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:34:22,238][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:34:22,737][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:34:23,237][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:34:23,736][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:34:24,233][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:34:24,731][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:34:25,229][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:34:25,728][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:34:26,228][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:34:26,727][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:34:27,227][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:34:27,725][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:34:28,224][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:34:28,723][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:34:29,224][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:34:29,724][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:34:30,224][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:34:30,724][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:34:31,225][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:34:31,724][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:34:32,224][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:34:32,724][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:34:33,223][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:34:33,723][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:34:34,223][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:34:34,723][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:34:35,224][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:34:35,724][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:34:36,223][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:34:36,721][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:34:37,220][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:34:37,718][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:34:38,215][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:34:38,712][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:34:39,210][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:34:39,707][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:34:40,206][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:34:40,705][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:34:41,203][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:34:41,702][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:34:42,201][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:34:42,700][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:34:43,200][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:34:43,699][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:34:44,198][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:34:44,698][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:34:45,198][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:34:45,697][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:34:46,195][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:34:46,692][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:34:47,190][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:34:47,688][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:34:48,188][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:34:48,689][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:34:49,188][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:34:49,685][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:34:50,183][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:34:50,680][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:34:51,178][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:34:51,677][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:34:52,175][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:34:52,676][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:34:53,176][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:34:53,677][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:34:54,176][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:34:54,676][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:34:55,176][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:34:55,677][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:34:56,177][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:34:56,677][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:34:57,176][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:34:57,673][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:34:58,172][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:34:58,670][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:34:59,168][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:34:59,666][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:35:00,166][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:35:00,665][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:35:01,165][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:35:01,663][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:35:02,163][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:35:02,662][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:35:03,162][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:35:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:35:04,161][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:35:04,661][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:35:05,160][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:35:05,659][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:35:06,159][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:35:06,658][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:35:07,158][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:35:07,658][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:35:08,158][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 05:35:08,797][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 05:35:09,537][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:35:09,539][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:35:09,541][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:35:10,196][__main__][INFO] - Iteration 426 took 1m 14s (9.34% Gen, 89.78% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 59m 11s. Estimated total time: 61h 59m 39s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 59s, 500 more iterations: 10h 19m 56s. [2026-03-26 05:35:10,199][__main__][INFO] - Starting iteration 426. [2026-03-26 05:35:10,599][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:35:10,600][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:35:11,716][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:35:17,608][__main__][INFO] - Number of regex retries in iteration 426: 1 [2026-03-26 05:35:17,609][__main__][INFO] - agents played in iteration 426 are Alice, Bob [2026-03-26 05:35:18,492][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:35:18,537][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:35:18,537][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:35:18,538][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:35:19,161][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:35:19,618][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:35:20,125][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:35:20,626][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:35:21,127][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:35:21,629][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:35:22,130][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:35:22,631][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:35:23,132][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:35:23,633][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:35:24,133][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:35:24,632][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:35:25,132][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:35:25,632][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:35:26,132][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:35:26,632][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:35:27,133][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:35:27,634][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:35:28,134][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:35:28,634][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:35:29,133][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:35:29,632][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:35:30,131][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:35:30,631][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:35:31,131][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:35:31,630][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:35:32,128][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:35:32,628][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:35:33,126][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:35:33,626][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:35:34,125][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:35:34,624][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:35:35,123][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:35:35,622][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:35:36,122][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:35:36,620][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:35:37,121][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:35:37,620][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:35:38,119][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:35:38,619][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:35:39,118][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:35:39,619][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:35:40,120][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:35:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:35:41,122][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:35:41,622][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:35:42,124][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:35:42,625][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:35:43,126][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:35:43,626][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:35:44,125][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:35:44,624][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:35:45,124][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:35:45,627][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:35:46,128][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:35:46,627][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:35:47,127][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:35:47,627][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:35:48,128][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:35:48,630][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:35:49,130][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:35:49,631][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:35:50,130][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:35:50,629][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:35:51,129][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:35:51,631][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:35:52,131][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:35:52,631][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:35:53,130][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:35:53,631][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:35:54,132][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:35:54,633][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:35:55,135][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:35:55,634][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:35:56,133][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:35:56,632][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:35:57,131][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:35:57,631][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:35:58,133][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:35:58,633][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:35:59,132][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:35:59,632][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:36:00,131][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:36:00,630][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:36:01,130][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:36:01,630][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:36:02,129][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:36:02,629][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:36:03,129][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:36:03,628][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:36:04,127][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:36:04,626][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:36:05,125][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:36:05,624][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:36:06,124][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:36:06,623][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:36:07,122][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:36:07,620][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:36:08,119][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:36:08,619][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:36:09,119][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:36:09,618][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:36:10,118][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:36:10,617][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:36:11,117][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:36:11,616][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:36:12,117][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:36:12,618][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:36:13,119][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:36:13,619][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:36:14,119][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:36:14,619][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:36:15,119][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:36:15,619][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:36:16,119][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:36:16,618][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:36:17,117][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:36:17,617][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:36:18,116][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:36:18,618][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:36:19,118][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:36:19,618][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:36:20,118][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:36:20,618][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:36:21,117][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:36:21,617][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:36:22,116][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:36:22,617][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:36:23,117][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21750 tokens. [2026-03-26 05:36:23,769][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 05:36:24,554][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:36:24,556][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:36:24,558][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:36:25,219][__main__][INFO] - Iteration 427 took 1m 14s (9.39% Gen, 89.72% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 9m 18s. Estimated total time: 62h 11m 1s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 22s, 500 more iterations: 10h 21m 50s. [2026-03-26 05:36:25,221][__main__][INFO] - Starting iteration 427. [2026-03-26 05:36:25,618][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:36:25,619][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:36:26,784][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:36:32,632][__main__][INFO] - Number of regex retries in iteration 427: 1 [2026-03-26 05:36:32,633][__main__][INFO] - agents played in iteration 427 are Alice, Bob [2026-03-26 05:36:33,516][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:36:33,561][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:36:33,562][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:36:33,562][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:36:34,172][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:36:34,627][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:36:35,130][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:36:35,629][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:36:36,127][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:36:36,626][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:36:37,125][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:36:37,624][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:36:38,124][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:36:38,624][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:36:39,123][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:36:39,622][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:36:40,121][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:36:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:36:41,120][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:36:41,619][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:36:42,118][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:36:42,616][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:36:43,114][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:36:43,612][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:36:44,110][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:36:44,607][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:36:45,106][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:36:45,604][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:36:46,103][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:36:46,606][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:36:47,107][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:36:47,609][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:36:48,111][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:36:48,614][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:36:49,115][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:36:49,615][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:36:50,116][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:36:50,615][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:36:51,115][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:36:51,614][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:36:52,113][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:36:52,612][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:36:53,110][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:36:53,609][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:36:54,108][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:36:54,607][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:36:55,108][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:36:55,608][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:36:56,107][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:36:56,607][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:36:57,106][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:36:57,605][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:36:58,105][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:36:58,604][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:36:59,102][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:36:59,600][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:37:00,098][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:37:00,595][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:37:01,093][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:37:01,590][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:37:02,087][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:37:02,586][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:37:03,085][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:37:03,584][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:37:04,083][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:37:04,581][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:37:05,080][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:37:05,580][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:37:06,079][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:37:06,580][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:37:07,080][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:37:07,579][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:37:08,078][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:37:08,578][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:37:09,077][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:37:09,577][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:37:10,077][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:37:10,577][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:37:11,076][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:37:11,575][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:37:12,073][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:37:12,572][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:37:13,070][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:37:13,568][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:37:14,067][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:37:14,565][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:37:15,063][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:37:15,564][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:37:16,064][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:37:16,563][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:37:17,061][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:37:17,560][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:37:18,059][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:37:18,561][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:37:19,062][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:37:19,562][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:37:20,062][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:37:20,563][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:37:21,063][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:37:21,563][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:37:22,062][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:37:22,562][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:37:23,060][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:37:23,559][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:37:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:37:24,555][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:37:25,054][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:37:25,553][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:37:26,051][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:37:26,549][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:37:27,048][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:37:27,546][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:37:28,045][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:37:28,543][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:37:29,042][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:37:29,542][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:37:30,042][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:37:30,540][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:37:31,038][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:37:31,536][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:37:32,034][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:37:32,531][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:37:33,028][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:37:33,526][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:37:34,025][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:37:34,524][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:37:35,023][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:37:35,521][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:37:36,020][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:37:36,518][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:37:37,016][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:37:37,516][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:37:38,015][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 05:37:38,681][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:37:39,422][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:37:39,424][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:37:39,425][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:37:40,116][__main__][INFO] - Iteration 428 took 1m 14s (9.42% Gen, 89.66% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 1m 56s. Estimated total time: 62h 4m 54s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 49s. [2026-03-26 05:37:40,118][__main__][INFO] - Starting iteration 428. [2026-03-26 05:37:40,517][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:37:40,518][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:37:45,334][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:37:47,008][__main__][INFO] - Number of regex retries in iteration 428: 1 [2026-03-26 05:37:47,009][__main__][INFO] - agents played in iteration 428 are Alice, Bob [2026-03-26 05:37:47,899][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:37:47,944][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:37:47,945][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:37:47,946][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:37:48,559][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:37:49,012][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:37:49,516][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:37:50,014][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:37:50,511][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:37:51,010][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:37:51,509][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:37:52,007][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:37:52,506][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:37:53,007][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:37:53,507][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:37:54,006][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:37:54,505][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:37:55,006][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:37:55,505][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:37:56,005][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:37:56,503][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:37:57,003][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:37:57,502][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:37:58,001][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:37:58,502][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:37:59,000][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:37:59,500][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:37:59,999][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:38:00,498][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:38:00,999][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:38:01,499][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:38:01,999][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:38:02,500][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:38:02,999][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:38:03,499][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:38:03,998][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:38:04,499][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:38:05,018][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:38:05,522][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:38:06,021][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:38:06,522][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:38:07,022][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:38:07,521][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:38:08,021][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:38:08,521][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:38:09,021][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:38:09,521][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:38:10,020][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:38:10,520][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:38:11,019][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:38:11,519][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:38:12,020][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:38:12,520][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:38:13,020][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:38:13,520][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:38:14,019][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:38:14,518][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:38:15,017][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:38:15,516][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:38:16,017][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:38:16,517][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:38:17,017][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:38:17,516][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:38:18,016][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:38:18,516][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:38:19,015][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:38:19,514][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:38:20,013][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:38:20,512][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:38:21,010][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:38:21,508][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:38:22,005][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:38:22,503][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:38:23,002][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:38:23,500][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:38:23,999][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:38:24,499][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:38:24,999][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:38:25,499][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:38:25,998][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:38:26,496][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:38:26,995][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:38:27,493][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:38:27,991][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:38:28,490][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:38:28,988][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:38:29,487][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:38:29,984][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:38:30,483][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:38:30,984][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:38:31,482][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:38:31,983][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:38:32,482][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:38:32,981][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:38:33,478][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:38:33,976][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:38:34,474][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:38:34,972][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:38:35,469][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:38:35,968][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:38:36,467][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:38:36,967][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:38:37,467][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:38:37,967][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:38:38,466][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:38:38,966][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:38:39,466][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:38:39,966][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:38:40,466][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:38:40,965][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:38:41,464][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:38:41,963][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:38:42,462][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:38:42,962][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:38:43,461][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:38:43,961][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:38:44,461][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:38:44,959][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:38:45,458][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:38:45,958][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:38:46,457][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:38:46,955][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:38:47,453][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:38:47,952][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:38:48,453][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:38:48,951][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:38:49,449][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:38:49,947][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:38:50,446][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:38:50,944][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:38:51,442][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:38:51,941][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:38:52,440][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 05:38:53,078][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 05:38:53,856][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:38:53,859][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:38:53,860][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:38:54,719][__main__][INFO] - Iteration 429 took 1m 14s (8.75% Gen, 90.09% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 45m 56s. Estimated total time: 61h 50m 8s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 40s, 500 more iterations: 10h 18m 21s. [2026-03-26 05:38:54,722][__main__][INFO] - Starting iteration 429. [2026-03-26 05:38:55,119][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:38:55,120][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:39:02,201][__main__][INFO] - Number of regex retries in iteration 429: 0 [2026-03-26 05:39:02,202][__main__][INFO] - agents played in iteration 429 are Alice, Bob [2026-03-26 05:39:03,107][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:39:03,152][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:39:03,153][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:39:03,153][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:39:03,760][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:39:04,213][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:39:04,716][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:39:05,213][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:39:05,712][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:39:06,211][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:39:06,710][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:39:07,209][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:39:07,708][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:39:08,209][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:39:08,710][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:39:09,210][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:39:09,710][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:39:10,211][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:39:10,712][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:39:11,211][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:39:11,711][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:39:12,210][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:39:12,710][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:39:13,209][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:39:13,708][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:39:14,208][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:39:14,708][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:39:15,207][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:39:15,706][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:39:16,205][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:39:16,703][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:39:17,202][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:39:17,701][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:39:18,201][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:39:18,701][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:39:19,199][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:39:19,697][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:39:20,192][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:39:20,686][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:39:21,181][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:39:21,677][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:39:22,173][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:39:22,667][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:39:23,163][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:39:23,658][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:39:24,157][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:39:24,655][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:39:25,154][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:39:25,655][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:39:26,154][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:39:26,652][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:39:27,151][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:39:27,651][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:39:28,149][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:39:28,647][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:39:29,145][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:39:29,643][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:39:30,143][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:39:30,641][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:39:31,140][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:39:31,639][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:39:32,139][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:39:32,638][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:39:33,137][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:39:33,635][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:39:34,134][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:39:34,632][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:39:35,131][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:39:35,629][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:39:36,128][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:39:36,626][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:39:37,125][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:39:37,623][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:39:38,121][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:39:38,621][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:39:39,121][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:39:39,620][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:39:40,120][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:39:40,620][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:39:41,120][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:39:41,620][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:39:42,121][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:39:42,621][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:39:43,123][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:39:43,623][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:39:44,123][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:39:44,623][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:39:45,123][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:39:45,624][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:39:46,123][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:39:46,623][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:39:47,122][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:39:47,621][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:39:48,119][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:39:48,619][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:39:49,119][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:39:49,618][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:39:50,117][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:39:50,615][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:39:51,113][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:39:51,612][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:39:52,107][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:39:52,603][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:39:53,099][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:39:53,594][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:39:54,089][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:39:54,585][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:39:55,081][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:39:55,577][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:39:56,077][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:39:56,576][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:39:57,074][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:39:57,573][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:39:58,073][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:39:58,572][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:39:59,070][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:39:59,569][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:40:00,067][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:40:00,565][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:40:01,063][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:40:01,561][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:40:02,060][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:40:02,559][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:40:03,059][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:40:03,558][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:40:04,057][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:40:04,555][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:40:05,054][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:40:05,553][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:40:06,051][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:40:06,550][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:40:07,049][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:40:07,548][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 05:40:08,183][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 05:40:08,944][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:40:08,947][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:40:08,948][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:40:09,826][__main__][INFO] - Iteration 430 took 1m 14s (9.48% Gen, 89.34% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 9m 53s. Estimated total time: 62h 15m 21s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 30s, 500 more iterations: 10h 22m 33s. [2026-03-26 05:40:09,828][__main__][INFO] - Starting iteration 430. [2026-03-26 05:40:10,227][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:40:10,227][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:40:12,403][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:40:16,919][__main__][INFO] - Number of regex retries in iteration 430: 1 [2026-03-26 05:40:16,920][__main__][INFO] - agents played in iteration 430 are Alice, Bob [2026-03-26 05:40:17,795][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:40:17,840][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:40:17,841][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:40:17,841][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:40:18,450][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:40:18,905][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:40:19,409][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:40:19,909][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:40:20,408][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:40:20,909][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:40:21,409][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:40:21,909][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:40:22,409][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:40:22,908][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:40:23,407][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:40:23,908][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:40:24,408][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:40:24,909][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:40:25,410][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:40:25,909][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:40:26,408][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:40:26,908][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:40:27,407][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:40:27,907][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:40:28,406][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:40:28,905][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:40:29,404][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:40:29,904][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:40:30,404][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:40:30,904][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:40:31,403][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:40:31,903][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:40:32,402][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:40:32,903][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:40:33,402][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:40:33,903][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:40:34,402][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:40:34,902][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:40:35,401][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:40:35,900][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:40:36,400][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:40:36,898][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:40:37,397][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:40:37,896][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:40:38,394][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:40:38,894][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:40:39,394][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:40:39,894][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:40:40,393][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:40:40,892][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:40:41,392][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:40:41,891][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:40:42,390][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:40:42,889][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:40:43,390][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:40:43,889][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:40:44,388][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:40:44,886][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:40:45,386][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:40:45,886][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:40:46,386][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:40:46,886][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:40:47,385][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:40:47,885][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:40:48,387][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:40:48,887][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:40:49,387][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:40:49,887][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:40:50,387][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:40:50,886][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:40:51,386][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:40:51,886][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:40:52,385][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:40:52,885][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:40:53,384][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:40:53,884][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:40:54,384][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:40:54,883][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:40:55,383][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:40:55,884][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:40:56,383][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:40:56,884][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:40:57,383][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:40:57,883][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:40:58,383][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:40:58,882][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:40:59,381][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:40:59,880][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:41:00,381][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:41:00,881][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:41:01,380][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:41:01,879][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:41:02,377][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:41:02,876][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:41:03,375][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:41:03,873][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:41:04,372][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:41:04,871][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:41:05,370][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:41:05,869][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:41:06,368][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:41:06,867][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:41:07,366][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:41:07,865][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:41:08,364][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:41:08,863][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:41:09,361][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:41:09,860][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:41:10,358][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:41:10,859][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:41:11,359][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:41:11,859][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:41:12,359][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:41:12,859][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:41:13,359][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:41:13,858][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:41:14,358][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:41:14,857][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:41:15,356][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:41:15,855][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:41:16,354][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:41:16,853][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:41:17,352][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:41:17,852][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:41:18,354][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:41:18,853][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:41:19,352][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:41:19,851][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:41:20,350][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:41:20,849][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:41:21,348][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:41:21,847][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:41:22,346][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 05:41:22,978][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 05:41:23,934][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:41:23,936][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:41:23,938][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:41:24,703][__main__][INFO] - Iteration 431 took 1m 14s (8.99% Gen, 89.98% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 52h 57m 9s. Estimated total time: 62h 3m 51s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 7s, 500 more iterations: 10h 20m 38s. [2026-03-26 05:41:24,706][__main__][INFO] - Starting iteration 431. [2026-03-26 05:41:25,104][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:41:25,104][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:41:27,403][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:41:32,101][__main__][INFO] - Number of regex retries in iteration 431: 1 [2026-03-26 05:41:32,102][__main__][INFO] - agents played in iteration 431 are Alice, Bob [2026-03-26 05:41:32,989][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:41:33,034][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:41:33,034][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:41:33,035][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:41:33,660][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:41:34,117][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:41:34,622][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:41:35,123][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:41:35,623][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:41:36,123][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:41:36,623][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:41:37,124][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:41:37,623][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:41:38,125][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:41:38,626][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:41:39,126][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:41:39,625][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:41:40,124][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:41:40,625][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:41:41,124][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:41:41,624][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:41:42,124][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:41:42,623][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:41:43,122][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:41:43,621][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:41:44,121][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:41:44,621][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:41:45,120][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:41:45,624][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:41:46,126][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:41:46,627][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:41:47,127][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:41:47,626][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:41:48,128][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:41:48,630][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:41:49,130][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:41:49,630][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:41:50,130][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:41:50,630][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:41:51,130][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:41:51,630][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:41:52,130][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:41:52,630][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:41:53,129][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:41:53,630][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:41:54,130][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:41:54,629][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:41:55,129][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:41:55,629][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:41:56,128][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:41:56,628][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:41:57,126][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:41:57,627][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:41:58,132][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:41:58,633][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:41:59,136][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:41:59,638][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:42:00,140][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:42:00,642][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:42:01,145][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:42:01,647][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:42:02,147][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:42:02,646][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:42:03,145][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:42:03,645][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:42:04,145][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:42:04,644][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:42:05,144][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:42:05,644][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:42:06,143][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:42:06,642][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:42:07,142][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:42:07,641][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:42:08,143][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:42:08,642][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:42:09,142][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:42:09,642][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:42:10,141][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:42:10,640][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:42:11,138][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:42:11,638][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:42:12,139][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:42:12,639][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:42:13,138][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:42:13,637][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:42:14,136][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:42:14,635][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:42:15,133][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:42:15,632][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:42:16,130][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:42:16,629][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:42:17,128][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:42:17,627][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:42:18,127][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:42:18,629][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:42:19,129][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:42:19,628][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:42:20,128][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:42:20,627][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:42:21,127][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:42:21,627][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:42:22,127][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:42:22,627][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:42:23,127][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:42:23,627][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:42:24,127][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:42:24,628][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:42:25,127][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:42:25,628][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:42:26,127][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:42:26,627][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:42:27,127][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:42:27,626][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:42:28,126][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:42:28,625][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:42:29,124][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:42:29,623][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:42:30,126][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:42:30,628][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:42:31,131][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:42:31,631][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:42:32,133][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:42:32,635][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:42:33,138][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:42:33,639][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:42:34,139][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:42:34,639][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:42:35,138][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:42:35,637][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:42:36,136][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:42:36,636][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:42:37,134][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:42:37,633][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21736 tokens. [2026-03-26 05:42:38,266][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.53%, ΔTime: 00:01:04 [2026-03-26 05:42:39,004][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:42:39,006][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:42:39,008][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:42:39,667][__main__][INFO] - Iteration 432 took 1m 14s (9.38% Gen, 89.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 0m 13s. Estimated total time: 62h 8m 10s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 16s, 500 more iterations: 10h 21m 21s. [2026-03-26 05:42:39,669][__main__][INFO] - Starting iteration 432. [2026-03-26 05:42:40,068][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:42:40,069][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:42:40,652][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:42:47,488][__main__][INFO] - Number of regex retries in iteration 432: 1 [2026-03-26 05:42:47,489][__main__][INFO] - agents played in iteration 432 are Alice, Bob [2026-03-26 05:42:48,389][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:42:48,434][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:42:48,434][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:42:48,435][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:42:49,039][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:42:49,494][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:42:49,999][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:42:50,499][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:42:50,998][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:42:51,498][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:42:51,998][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:42:52,496][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:42:52,994][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:42:53,494][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:42:53,993][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:42:54,491][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:42:54,991][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:42:55,490][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:42:55,990][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:42:56,489][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:42:56,989][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:42:57,489][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:42:57,989][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:42:58,488][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:42:58,987][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:42:59,486][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:42:59,986][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:43:00,486][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:43:00,984][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:43:01,486][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:43:01,987][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:43:02,488][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:43:02,988][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:43:03,489][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:43:03,989][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:43:04,490][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:43:04,990][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:43:05,490][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:43:05,989][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:43:06,488][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:43:06,987][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:43:07,487][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:43:07,986][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:43:08,487][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:43:08,987][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:43:09,487][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:43:09,988][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:43:10,488][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:43:10,988][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:43:11,489][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:43:11,989][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:43:12,490][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:43:12,990][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:43:13,490][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:43:13,989][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:43:14,488][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:43:14,988][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:43:15,490][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:43:15,992][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:43:16,492][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:43:16,990][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:43:17,490][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:43:17,997][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:43:18,500][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:43:19,001][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:43:19,502][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:43:20,003][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:43:20,504][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:43:21,005][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:43:21,504][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:43:22,003][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:43:22,502][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:43:23,002][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:43:23,502][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:43:24,002][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:43:24,502][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:43:25,003][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:43:25,503][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:43:26,003][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:43:26,503][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:43:27,002][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:43:27,503][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:43:28,003][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:43:28,502][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:43:29,002][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:43:29,502][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:43:30,002][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:43:30,503][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:43:31,004][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:43:31,504][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:43:32,003][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:43:32,503][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:43:33,003][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:43:33,503][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:43:34,005][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:43:34,506][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:43:35,007][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:43:35,508][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:43:36,010][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:43:36,511][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:43:37,011][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:43:37,511][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:43:38,010][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:43:38,510][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:43:39,011][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:43:39,511][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:43:40,011][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:43:40,510][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:43:41,009][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:43:41,509][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:43:42,010][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:43:42,510][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:43:43,011][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:43:43,511][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:43:44,010][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:43:44,510][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:43:45,010][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:43:45,509][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:43:46,008][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:43:46,508][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:43:47,007][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:43:47,507][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:43:48,008][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:43:48,509][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:43:49,009][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:43:49,511][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:43:50,012][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:43:50,512][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:43:51,011][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:43:51,512][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:43:52,011][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:43:52,511][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:43:53,012][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21740 tokens. [2026-03-26 05:43:53,645][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:43:54,383][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:43:54,385][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:43:54,387][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:43:55,046][__main__][INFO] - Iteration 433 took 1m 14s (9.90% Gen, 89.22% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 19m 42s. Estimated total time: 62h 28m 55s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 57s, 500 more iterations: 10h 24m 49s. [2026-03-26 05:43:55,049][__main__][INFO] - Starting iteration 433. [2026-03-26 05:43:55,449][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:43:55,450][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:43:56,043][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:43:58,310][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:44:02,438][__main__][INFO] - Number of regex retries in iteration 433: 2 [2026-03-26 05:44:02,438][__main__][INFO] - agents played in iteration 433 are Alice, Bob [2026-03-26 05:44:03,340][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:44:03,385][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:44:03,386][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:44:03,386][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:44:03,995][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:44:04,449][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:44:04,952][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:44:05,452][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:44:05,951][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:44:06,450][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:44:06,951][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:44:07,451][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:44:07,949][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:44:08,448][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:44:08,946][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:44:09,445][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:44:09,944][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:44:10,443][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:44:10,942][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:44:11,441][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:44:11,939][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:44:12,437][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:44:12,935][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:44:13,433][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:44:13,931][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:44:14,429][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:44:14,928][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:44:15,427][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:44:15,925][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:44:16,425][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:44:16,926][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:44:17,425][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:44:17,925][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:44:18,426][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:44:18,926][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:44:19,426][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:44:19,924][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:44:20,422][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:44:20,921][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:44:21,420][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:44:21,919][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:44:22,416][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:44:22,916][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:44:23,414][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:44:23,912][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:44:24,410][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:44:24,909][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:44:25,409][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:44:25,908][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:44:26,407][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:44:26,906][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:44:27,404][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:44:27,904][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:44:28,403][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:44:28,902][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:44:29,401][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:44:29,901][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:44:30,400][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:44:30,899][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:44:31,396][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:44:31,894][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:44:32,392][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:44:32,890][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:44:33,389][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:44:33,888][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:44:34,386][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:44:34,885][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:44:35,383][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:44:35,882][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:44:36,380][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:44:36,878][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:44:37,378][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:44:37,878][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:44:38,376][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:44:38,876][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:44:39,374][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:44:39,873][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:44:40,371][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:44:40,869][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:44:41,368][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:44:41,867][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:44:42,365][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:44:42,864][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:44:43,362][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:44:43,861][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:44:44,358][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:44:44,856][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:44:45,356][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:44:45,857][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:44:46,356][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:44:46,855][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:44:47,353][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:44:47,852][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:44:48,353][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:44:48,853][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:44:49,353][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:44:49,852][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:44:50,350][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:44:50,849][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:44:51,349][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:44:51,849][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:44:52,347][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:44:52,846][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:44:53,345][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:44:53,844][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:44:54,342][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:44:54,841][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:44:55,339][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:44:55,837][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:44:56,336][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:44:56,835][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:44:57,334][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:44:57,833][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:44:58,332][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:44:58,831][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:44:59,331][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:44:59,829][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:45:00,328][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:45:00,828][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:45:01,327][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:45:01,825][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:45:02,324][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:45:02,823][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:45:03,321][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:45:03,819][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:45:04,317][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:45:04,816][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:45:05,315][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:45:05,814][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:45:06,313][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:45:06,812][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:45:07,311][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:45:07,810][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21737 tokens. [2026-03-26 05:45:08,443][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 05:45:09,180][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:45:09,182][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:45:09,184][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:45:09,834][__main__][INFO] - Iteration 434 took 1m 14s (9.39% Gen, 89.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 48m 48s. Estimated total time: 61h 59m 16s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 58s, 500 more iterations: 10h 19m 52s. [2026-03-26 05:45:09,837][__main__][INFO] - Starting iteration 434. [2026-03-26 05:45:10,237][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:45:10,238][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:45:16,772][__main__][INFO] - Number of regex retries in iteration 434: 0 [2026-03-26 05:45:16,773][__main__][INFO] - agents played in iteration 434 are Alice, Bob [2026-03-26 05:45:17,651][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:45:17,696][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:45:17,696][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:45:17,697][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:45:18,306][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:45:18,764][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:45:19,267][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:45:19,766][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:45:20,266][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:45:20,765][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:45:21,264][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:45:21,763][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:45:22,263][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:45:22,762][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:45:23,262][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:45:23,760][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:45:24,259][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:45:24,758][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:45:25,258][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:45:25,757][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:45:26,255][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:45:26,754][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:45:27,252][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:45:27,750][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:45:28,248][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:45:28,747][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:45:29,245][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:45:29,743][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:45:30,242][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:45:30,739][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:45:31,235][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:45:31,731][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:45:32,229][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:45:32,726][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:45:33,222][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:45:33,718][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:45:34,215][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:45:34,715][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:45:35,214][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:45:35,713][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:45:36,212][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:45:36,711][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:45:37,209][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:45:37,708][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:45:38,209][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:45:38,710][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:45:39,209][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:45:39,709][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:45:40,208][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:45:40,707][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:45:41,207][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:45:41,706][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:45:42,206][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:45:42,705][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:45:43,203][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:45:43,702][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:45:44,200][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:45:44,697][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:45:45,194][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:45:45,692][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:45:46,190][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:45:46,690][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:45:47,190][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:45:47,688][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:45:48,189][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:45:48,689][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:45:49,187][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:45:49,686][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:45:50,185][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:45:50,683][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:45:51,183][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:45:51,681][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:45:52,179][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:45:52,678][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:45:53,176][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:45:53,674][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:45:54,173][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:45:54,671][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:45:55,170][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:45:55,669][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:45:56,167][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:45:56,666][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:45:57,164][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:45:57,662][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:45:58,164][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:45:58,663][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:45:59,162][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:45:59,661][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:46:00,159][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:46:00,657][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:46:01,155][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:46:01,654][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:46:02,152][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:46:02,647][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:46:03,145][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:46:03,642][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:46:04,138][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:46:04,635][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:46:05,131][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:46:05,628][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:46:06,124][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:46:06,623][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:46:07,124][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:46:07,623][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:46:08,123][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:46:08,622][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:46:09,120][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:46:09,619][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:46:10,118][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:46:10,617][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:46:11,115][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:46:11,614][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:46:12,112][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:46:12,611][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:46:13,109][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:46:13,609][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:46:14,107][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:46:14,606][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:46:15,104][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:46:15,606][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:46:16,107][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:46:16,605][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:46:17,104][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:46:17,604][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:46:18,104][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:46:18,603][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:46:19,103][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:46:19,603][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:46:20,102][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:46:20,600][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:46:21,099][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:46:21,597][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:46:22,096][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21737 tokens. [2026-03-26 05:46:22,722][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 05:46:23,480][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:46:23,482][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:46:23,484][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:46:24,161][__main__][INFO] - Iteration 435 took 1m 13s (8.84% Gen, 90.24% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 24m 30s. Estimated total time: 61h 36m 12s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 12s, 500 more iterations: 10h 16m 2s. [2026-03-26 05:46:24,163][__main__][INFO] - Starting iteration 435. [2026-03-26 05:46:24,560][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:46:24,561][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:46:29,810][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:46:30,883][__main__][INFO] - Number of regex retries in iteration 435: 1 [2026-03-26 05:46:30,884][__main__][INFO] - agents played in iteration 435 are Alice, Bob [2026-03-26 05:46:31,770][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:46:31,815][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:46:31,816][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:46:31,816][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:46:32,421][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:46:32,880][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:46:33,384][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:46:33,884][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:46:34,384][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:46:34,883][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:46:35,383][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:46:35,882][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:46:36,385][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:46:36,884][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:46:37,383][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:46:37,881][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:46:38,379][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:46:38,877][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:46:39,374][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:46:39,874][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:46:40,373][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:46:40,872][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:46:41,371][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:46:41,870][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:46:42,368][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:46:42,867][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:46:43,366][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:46:43,866][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:46:44,365][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:46:44,865][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:46:45,365][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:46:45,864][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:46:46,363][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:46:46,862][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:46:47,361][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:46:47,861][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:46:48,364][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:46:48,864][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:46:49,363][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:46:49,862][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:46:50,361][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:46:50,860][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:46:51,359][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:46:51,859][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:46:52,358][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:46:52,857][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:46:53,355][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:46:53,853][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:46:54,351][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:46:54,849][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:46:55,349][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:46:55,847][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:46:56,346][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:46:56,844][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:46:57,342][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:46:57,840][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:46:58,338][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:46:58,835][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:46:59,334][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:46:59,831][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:47:00,329][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:47:00,828][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:47:01,326][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:47:01,825][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:47:02,324][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:47:02,822][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:47:03,319][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:47:03,818][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:47:04,316][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:47:04,815][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:47:05,316][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:47:05,815][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:47:06,315][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:47:06,815][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:47:07,314][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:47:07,815][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:47:08,315][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:47:08,814][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:47:09,312][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:47:09,810][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:47:10,309][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:47:10,807][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:47:11,306][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:47:11,805][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:47:12,304][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:47:12,802][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:47:13,302][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:47:13,800][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:47:14,298][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:47:14,795][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:47:15,293][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:47:15,791][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:47:16,290][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:47:16,789][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:47:17,288][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:47:17,787][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:47:18,288][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:47:18,788][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:47:19,287][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:47:19,786][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:47:20,286][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:47:20,785][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:47:21,284][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:47:21,784][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:47:22,284][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:47:22,784][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:47:23,285][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:47:23,784][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:47:24,284][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:47:24,783][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:47:25,282][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:47:25,780][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:47:26,278][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:47:26,777][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:47:27,276][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:47:27,774][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:47:28,273][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:47:28,771][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:47:29,268][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:47:29,766][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:47:30,264][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:47:30,763][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:47:31,261][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:47:31,759][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:47:32,256][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:47:32,755][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:47:33,253][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:47:33,750][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:47:34,249][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:47:34,747][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:47:35,246][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:47:35,744][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:47:36,243][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 05:47:36,871][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 05:47:37,607][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:47:37,610][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:47:37,611][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:47:38,311][__main__][INFO] - Iteration 436 took 1m 13s (8.57% Gen, 90.48% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 14m 38s. Estimated total time: 61h 27m 34s. Time estimates for 10 more iterations: 12m 17s, 100 more iterations: 2h 2m 55s, 500 more iterations: 10h 14m 35s. [2026-03-26 05:47:38,314][__main__][INFO] - Starting iteration 436. [2026-03-26 05:47:38,711][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:47:38,711][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:47:42,272][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:47:45,784][__main__][INFO] - Number of regex retries in iteration 436: 1 [2026-03-26 05:47:45,785][__main__][INFO] - agents played in iteration 436 are Alice, Bob [2026-03-26 05:47:46,679][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:47:46,724][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:47:46,725][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:47:46,725][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:47:47,335][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:47:47,791][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:47:48,296][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:47:48,798][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:47:49,298][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:47:49,799][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:47:50,299][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:47:50,800][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:47:51,301][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:47:51,802][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:47:52,301][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:47:52,800][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:47:53,298][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:47:53,798][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:47:54,297][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:47:54,796][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:47:55,295][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:47:55,794][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:47:56,293][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:47:56,792][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:47:57,290][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:47:57,789][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:47:58,290][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:47:58,790][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:47:59,289][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:47:59,787][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:48:00,286][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:48:00,786][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:48:01,284][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:48:01,784][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:48:02,283][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:48:02,782][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:48:03,281][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:48:03,780][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:48:04,279][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:48:04,778][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:48:05,278][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:48:05,777][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:48:06,276][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:48:06,775][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:48:07,274][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:48:07,772][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:48:08,271][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:48:08,769][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:48:09,267][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:48:09,765][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:48:10,264][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:48:10,763][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:48:11,262][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:48:11,760][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:48:12,258][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:48:12,756][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:48:13,255][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:48:13,753][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:48:14,252][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:48:14,751][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:48:15,249][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:48:15,748][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:48:16,247][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:48:16,746][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:48:17,244][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:48:17,744][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:48:18,246][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:48:18,746][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:48:19,244][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:48:19,746][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:48:20,247][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:48:20,748][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:48:21,248][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:48:21,749][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:48:22,250][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:48:22,750][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:48:23,250][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:48:23,749][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:48:24,248][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:48:24,747][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:48:25,246][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:48:25,746][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:48:26,244][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:48:26,743][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:48:27,242][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:48:27,740][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:48:28,239][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:48:28,737][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:48:29,235][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:48:29,734][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:48:30,233][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:48:30,731][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:48:31,230][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:48:31,729][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:48:32,227][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:48:32,725][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:48:33,224][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:48:33,723][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:48:34,221][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:48:34,719][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:48:35,218][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:48:35,716][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:48:36,214][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:48:36,714][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:48:37,213][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:48:37,713][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:48:38,211][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:48:38,710][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:48:39,209][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:48:39,708][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:48:40,206][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:48:40,705][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:48:41,203][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:48:41,702][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:48:42,200][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:48:42,698][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:48:43,196][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:48:43,694][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:48:44,192][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:48:44,692][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:48:45,192][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:48:45,690][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:48:46,189][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:48:46,688][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:48:47,188][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:48:47,687][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:48:48,190][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:48:48,691][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:48:49,190][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:48:49,689][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:48:50,188][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:48:50,687][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:48:51,186][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-26 05:48:51,819][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 05:48:52,556][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:48:52,558][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:48:52,560][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:48:53,266][__main__][INFO] - Iteration 437 took 1m 14s (9.49% Gen, 89.56% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 53m 36s. Estimated total time: 62h 7m 47s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 15s, 500 more iterations: 10h 21m 17s. [2026-03-26 05:48:53,268][__main__][INFO] - Starting iteration 437. [2026-03-26 05:48:53,666][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:48:53,667][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:49:00,280][__main__][INFO] - Number of regex retries in iteration 437: 0 [2026-03-26 05:49:00,281][__main__][INFO] - agents played in iteration 437 are Alice, Bob [2026-03-26 05:49:01,179][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:49:01,224][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:49:01,225][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:49:01,225][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:49:01,836][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:49:02,291][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:49:02,794][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:49:03,292][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:49:03,791][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:49:04,290][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:49:04,790][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:49:05,289][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:49:05,788][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:49:06,289][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:49:06,790][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:49:07,290][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:49:07,790][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:49:08,291][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:49:08,791][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:49:09,291][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:49:09,791][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:49:10,289][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:49:10,789][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:49:11,288][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:49:11,787][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:49:12,286][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:49:12,785][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:49:13,285][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:49:13,784][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:49:14,286][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:49:14,786][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:49:15,289][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:49:15,794][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:49:16,295][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:49:16,795][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:49:17,295][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:49:17,796][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:49:18,298][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:49:18,798][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:49:19,298][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:49:19,798][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:49:20,297][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:49:20,796][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:49:21,295][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:49:21,795][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:49:22,293][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:49:22,793][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:49:23,292][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:49:23,791][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:49:24,289][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:49:24,788][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:49:25,288][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:49:25,786][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:49:26,286][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:49:26,786][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:49:27,285][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:49:27,786][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:49:28,286][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:49:28,786][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:49:29,286][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:49:29,785][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:49:30,284][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:49:30,783][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:49:31,281][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:49:31,780][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:49:32,277][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:49:32,774][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:49:33,273][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:49:33,772][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:49:34,270][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:49:34,768][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:49:35,267][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:49:35,766][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:49:36,264][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:49:36,763][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:49:37,262][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:49:37,762][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:49:38,262][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:49:38,761][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:49:39,261][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:49:39,760][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:49:40,259][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:49:40,758][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:49:41,257][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:49:41,757][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:49:42,255][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:49:42,752][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:49:43,252][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:49:43,750][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:49:44,247][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:49:44,746][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:49:45,245][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:49:45,743][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:49:46,243][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:49:46,743][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:49:47,244][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:49:47,745][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:49:48,247][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:49:48,749][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:49:49,250][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:49:49,751][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:49:50,251][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:49:50,750][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:49:51,250][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:49:51,750][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:49:52,250][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:49:52,750][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:49:53,250][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:49:53,750][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:49:54,248][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:49:54,747][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:49:55,247][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:49:55,745][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:49:56,244][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:49:56,743][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:49:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:49:57,741][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:49:58,242][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:49:58,740][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:49:59,239][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:49:59,738][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:50:00,237][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:50:00,735][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:50:01,234][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:50:01,733][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:50:02,232][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:50:02,730][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:50:03,229][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:50:03,728][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:50:04,227][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:50:04,726][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:50:05,225][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:50:05,724][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-26 05:50:06,355][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 05:50:07,109][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:50:07,111][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:50:07,112][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:50:07,777][__main__][INFO] - Iteration 438 took 1m 14s (8.92% Gen, 90.18% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 30m 7s. Estimated total time: 61h 45m 33s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 31s, 500 more iterations: 10h 17m 35s. [2026-03-26 05:50:07,779][__main__][INFO] - Starting iteration 438. [2026-03-26 05:50:08,179][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:50:08,180][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:50:08,769][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:50:09,850][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:50:15,150][__main__][INFO] - Number of regex retries in iteration 438: 2 [2026-03-26 05:50:15,151][__main__][INFO] - agents played in iteration 438 are Alice, Bob [2026-03-26 05:50:16,042][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:50:16,087][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:50:16,088][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:50:16,088][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:50:16,692][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:50:17,148][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:50:17,652][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:50:18,154][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:50:18,654][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:50:19,153][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:50:19,653][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:50:20,152][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:50:20,652][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:50:21,150][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:50:21,649][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:50:22,148][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:50:22,647][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:50:23,146][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:50:23,645][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:50:24,144][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:50:24,642][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:50:25,144][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:50:25,643][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:50:26,142][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:50:26,642][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:50:27,140][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:50:27,638][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:50:28,137][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:50:28,636][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:50:29,136][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:50:29,635][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:50:30,136][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:50:30,636][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:50:31,135][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:50:31,636][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:50:32,136][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:50:32,636][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:50:33,137][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:50:33,637][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:50:34,138][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:50:34,638][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:50:35,138][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:50:35,639][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:50:36,139][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:50:36,639][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:50:37,138][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:50:37,637][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:50:38,135][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:50:38,634][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:50:39,133][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:50:39,632][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:50:40,131][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:50:40,630][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:50:41,130][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:50:41,630][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:50:42,129][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:50:42,627][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:50:43,126][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:50:43,627][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:50:44,127][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:50:44,627][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:50:45,127][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:50:45,626][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:50:46,125][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:50:46,627][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:50:47,129][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:50:47,628][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:50:48,129][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:50:48,631][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:50:49,131][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:50:49,631][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:50:50,130][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:50:50,630][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:50:51,130][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:50:51,630][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:50:52,130][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:50:52,628][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:50:53,126][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:50:53,625][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:50:54,124][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:50:54,623][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:50:55,123][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:50:55,622][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:50:56,120][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:50:56,618][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:50:57,116][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:50:57,615][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:50:58,114][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:50:58,614][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:50:59,113][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:50:59,612][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:51:00,112][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:51:00,610][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:51:01,110][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:51:01,611][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:51:02,111][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:51:02,610][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:51:03,111][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:51:03,611][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:51:04,111][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:51:04,611][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:51:05,113][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:51:05,614][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:51:06,114][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:51:06,615][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:51:07,115][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:51:07,614][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:51:08,113][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:51:08,614][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:51:09,113][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:51:09,613][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:51:10,112][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:51:10,611][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:51:11,110][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:51:11,608][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:51:12,107][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:51:12,606][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:51:13,105][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:51:13,605][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:51:14,104][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:51:14,602][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:51:15,099][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:51:15,596][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:51:16,094][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:51:16,593][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:51:17,092][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:51:17,591][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:51:18,091][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:51:18,593][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:51:19,092][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:51:19,591][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:51:20,090][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:51:20,591][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 05:51:21,227][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 05:51:21,957][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:51:21,960][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:51:21,961][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:51:22,650][__main__][INFO] - Iteration 439 took 1m 14s (9.36% Gen, 89.71% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 46m 54s. Estimated total time: 62h 3m 34s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 7s, 500 more iterations: 10h 20m 35s. [2026-03-26 05:51:22,652][__main__][INFO] - Starting iteration 439. [2026-03-26 05:51:23,060][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:51:23,060][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:51:25,305][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:51:29,941][__main__][INFO] - Number of regex retries in iteration 439: 1 [2026-03-26 05:51:29,942][__main__][INFO] - agents played in iteration 439 are Alice, Bob [2026-03-26 05:51:30,831][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:51:30,876][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:51:30,877][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:51:30,877][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:51:31,486][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:51:31,942][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:51:32,445][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:51:32,945][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:51:33,444][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:51:33,944][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:51:34,445][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:51:34,946][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:51:35,445][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:51:35,944][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:51:36,443][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:51:36,942][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:51:37,439][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:51:37,937][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:51:38,437][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:51:38,936][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:51:39,435][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:51:39,936][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:51:40,435][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:51:40,934][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:51:41,433][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:51:41,933][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:51:42,433][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:51:42,932][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:51:43,432][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:51:43,933][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:51:44,434][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:51:44,934][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:51:45,434][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:51:45,934][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:51:46,435][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:51:46,935][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:51:47,435][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:51:47,935][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:51:48,437][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:51:48,937][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:51:49,436][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:51:49,936][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:51:50,436][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:51:50,934][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:51:51,433][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:51:51,933][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:51:52,433][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:51:52,932][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:51:53,432][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:51:53,932][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:51:54,432][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:51:54,930][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:51:55,432][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:51:55,930][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:51:56,430][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:51:56,929][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:51:57,428][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:51:57,928][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:51:58,426][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:51:58,925][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:51:59,423][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:51:59,922][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:52:00,422][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:52:00,921][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:52:01,419][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:52:01,918][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:52:02,416][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:52:02,914][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:52:03,412][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:52:03,912][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:52:04,412][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:52:04,913][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:52:05,412][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:52:05,912][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:52:06,412][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:52:06,912][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:52:07,412][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:52:07,912][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:52:08,411][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:52:08,910][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:52:09,409][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:52:09,908][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:52:10,408][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:52:10,907][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:52:11,406][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:52:11,906][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:52:12,406][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:52:12,904][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:52:13,403][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:52:13,902][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:52:14,400][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:52:14,899][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:52:15,399][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:52:15,902][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:52:16,402][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:52:16,902][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:52:17,401][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:52:17,902][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:52:18,404][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:52:18,906][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:52:19,406][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:52:19,905][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:52:20,403][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:52:20,901][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:52:21,399][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:52:21,896][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:52:22,394][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:52:22,893][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:52:23,392][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:52:23,891][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:52:24,391][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:52:24,890][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:52:25,391][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:52:25,891][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:52:26,391][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:52:26,892][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:52:27,392][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:52:27,891][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:52:28,389][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:52:28,889][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:52:29,388][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:52:29,887][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:52:30,386][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:52:30,885][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:52:31,383][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:52:31,882][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:52:32,380][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:52:32,879][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:52:33,378][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:52:33,877][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:52:34,376][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:52:34,874][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:52:35,373][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 05:52:36,005][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 05:52:36,737][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:52:36,740][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:52:36,741][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:52:37,392][__main__][INFO] - Iteration 440 took 1m 14s (9.26% Gen, 89.87% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 38m 42s. Estimated total time: 61h 56m 37s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 53s, 500 more iterations: 10h 19m 26s. [2026-03-26 05:52:37,394][__main__][INFO] - Starting iteration 440. [2026-03-26 05:52:37,792][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:52:37,793][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:52:39,362][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:52:42,651][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:52:44,710][__main__][INFO] - Number of regex retries in iteration 440: 2 [2026-03-26 05:52:44,711][__main__][INFO] - agents played in iteration 440 are Alice, Bob [2026-03-26 05:52:45,607][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:52:45,934][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:52:45,935][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:52:45,935][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:52:46,541][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:52:46,995][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:52:47,499][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:52:47,998][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:52:48,500][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:52:48,999][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:52:49,498][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:52:49,997][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:52:50,496][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:52:50,995][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:52:51,493][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:52:51,991][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:52:52,489][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:52:52,988][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:52:53,487][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:52:53,986][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:52:54,483][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:52:54,983][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:52:55,482][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:52:55,980][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:52:56,478][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:52:56,976][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:52:57,474][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:52:57,974][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:52:58,472][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:52:58,972][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:52:59,471][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:52:59,970][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:53:00,469][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:53:00,969][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:53:01,468][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:53:01,966][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:53:02,465][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:53:02,963][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:53:03,461][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:53:03,959][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:53:04,457][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:53:04,955][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:53:05,453][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:53:05,952][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:53:06,451][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:53:06,952][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:53:07,455][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:53:07,958][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:53:08,460][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:53:08,963][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:53:09,465][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:53:09,968][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:53:10,469][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:53:10,969][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:53:11,468][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:53:11,967][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:53:12,465][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:53:12,963][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:53:13,462][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:53:13,960][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:53:14,458][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:53:14,957][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:53:15,455][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:53:15,954][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:53:16,453][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:53:16,951][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:53:17,450][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:53:17,949][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:53:18,450][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:53:18,948][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:53:19,448][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:53:19,947][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:53:20,446][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:53:20,943][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:53:21,442][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:53:21,939][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:53:22,437][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:53:22,934][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:53:23,433][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:53:23,931][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:53:24,429][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:53:24,926][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:53:25,426][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:53:25,924][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:53:26,422][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:53:26,920][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:53:27,419][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:53:27,918][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:53:28,417][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:53:28,916][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:53:29,415][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:53:29,914][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:53:30,413][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:53:30,912][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:53:31,412][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:53:31,911][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:53:32,410][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:53:32,910][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:53:33,409][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:53:33,909][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:53:34,408][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:53:34,907][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:53:35,407][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:53:35,906][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:53:36,405][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:53:36,905][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:53:37,404][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:53:37,902][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:53:38,400][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:53:38,902][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:53:39,405][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:53:39,906][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:53:40,408][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:53:40,911][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:53:41,413][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:53:41,916][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:53:42,418][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:53:42,917][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:53:43,415][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:53:43,914][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:53:44,413][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:53:44,911][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:53:45,412][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:53:45,914][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:53:46,414][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:53:46,913][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:53:47,412][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:53:47,911][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:53:48,411][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:53:48,910][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:53:49,410][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:53:49,908][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:53:50,407][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 05:53:51,043][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 05:53:51,780][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:53:51,783][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:53:51,784][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:53:52,436][__main__][INFO] - Iteration 441 took 1m 14s (9.27% Gen, 89.86% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 52h 53m 1s. Estimated total time: 62h 12m 12s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 2s. [2026-03-26 05:53:52,438][__main__][INFO] - Starting iteration 441. [2026-03-26 05:53:52,841][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:53:52,842][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:53:53,956][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:53:53,958][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 books, 10 balls, 10 hats did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:54:00,204][__main__][INFO] - Number of regex retries in iteration 441: 2 [2026-03-26 05:54:00,205][__main__][INFO] - agents played in iteration 441 are Alice, Bob [2026-03-26 05:54:01,112][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:54:01,157][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:54:01,158][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:54:01,158][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:54:01,766][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:54:02,221][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:54:02,725][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:54:03,224][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:54:03,723][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:54:04,222][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:54:04,720][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:54:05,217][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:54:05,716][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:54:06,215][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:54:06,715][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:54:07,214][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:54:07,713][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:54:08,212][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:54:08,711][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:54:09,211][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:54:09,710][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:54:10,209][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:54:10,708][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:54:11,206][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:54:11,705][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:54:12,205][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:54:12,705][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:54:13,204][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:54:13,705][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:54:14,204][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:54:14,702][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:54:15,200][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:54:15,698][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:54:16,197][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:54:16,695][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:54:17,193][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:54:17,691][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:54:18,193][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:54:18,694][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:54:19,194][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:54:19,690][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:54:20,191][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:54:20,690][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:54:21,189][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:54:21,688][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:54:22,188][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:54:22,687][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:54:23,187][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:54:23,686][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:54:24,186][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:54:24,686][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:54:25,187][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:54:25,687][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:54:26,185][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:54:26,683][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:54:27,182][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:54:27,681][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:54:28,179][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:54:28,678][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:54:29,176][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:54:29,675][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:54:30,175][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:54:30,673][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:54:31,172][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:54:31,671][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:54:32,170][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:54:32,669][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:54:33,169][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:54:33,668][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:54:34,167][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:54:34,665][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:54:35,164][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:54:35,662][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:54:36,159][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:54:36,657][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:54:37,155][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:54:37,654][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:54:38,153][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:54:38,652][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:54:39,151][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:54:39,650][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:54:40,149][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:54:40,648][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:54:41,147][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:54:41,646][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:54:42,146][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:54:42,646][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:54:43,144][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:54:43,643][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:54:44,142][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:54:44,641][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:54:45,140][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:54:45,639][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:54:46,138][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:54:46,637][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:54:47,135][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:54:47,635][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:54:48,134][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:54:48,636][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:54:49,136][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:54:49,635][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:54:50,135][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:54:50,635][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:54:51,134][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:54:51,634][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:54:52,134][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:54:52,633][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:54:53,132][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:54:53,633][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:54:54,133][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:54:54,633][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:54:55,133][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:54:55,633][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:54:56,132][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:54:56,632][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:54:57,133][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:54:57,633][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:54:58,133][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:54:58,633][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:54:59,133][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:54:59,632][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:55:00,133][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:55:00,633][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:55:01,133][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:55:01,633][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:55:02,133][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:55:02,632][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:55:03,132][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:55:03,632][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:55:04,132][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:55:04,631][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:55:05,131][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:55:05,629][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21723 tokens. [2026-03-26 05:55:06,264][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-26 05:55:07,064][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:55:07,067][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:55:07,068][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:55:07,843][__main__][INFO] - Iteration 442 took 1m 15s (9.82% Gen, 89.15% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 9m 43s. Estimated total time: 62h 30m 9s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 0s, 500 more iterations: 10h 25m 1s. [2026-03-26 05:55:07,846][__main__][INFO] - Starting iteration 442. [2026-03-26 05:55:08,246][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:55:08,247][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:55:08,837][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:55:13,690][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:55:15,166][__main__][INFO] - Number of regex retries in iteration 442: 2 [2026-03-26 05:55:15,167][__main__][INFO] - agents played in iteration 442 are Alice, Bob [2026-03-26 05:55:16,046][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:55:16,371][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:55:16,372][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:55:16,372][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:55:16,988][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:55:17,444][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:55:17,949][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:55:18,450][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:55:18,950][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:55:19,450][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:55:19,949][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:55:20,448][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:55:20,948][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:55:21,448][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:55:21,947][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:55:22,447][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:55:22,947][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:55:23,446][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:55:23,945][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:55:24,444][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:55:24,943][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:55:25,441][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:55:25,940][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:55:26,439][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:55:26,938][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:55:27,437][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:55:27,935][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:55:28,434][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:55:28,933][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:55:29,435][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:55:29,935][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:55:30,436][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:55:30,936][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:55:31,437][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:55:31,937][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:55:32,437][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:55:32,938][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:55:33,437][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:55:33,937][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:55:34,436][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:55:34,936][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:55:35,435][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:55:35,935][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:55:36,435][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:55:36,934][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:55:37,433][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:55:37,933][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:55:38,433][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:55:38,932][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:55:39,432][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:55:39,932][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:55:40,430][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:55:40,930][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:55:41,429][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:55:41,930][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:55:42,430][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:55:42,929][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:55:43,428][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:55:43,929][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:55:44,429][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:55:44,929][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:55:45,428][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:55:45,927][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:55:46,427][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:55:46,927][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:55:47,427][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:55:47,928][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:55:48,430][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:55:48,930][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:55:49,429][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:55:49,930][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:55:50,430][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:55:50,930][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:55:51,429][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:55:51,931][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:55:52,430][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:55:52,930][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:55:53,429][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:55:53,928][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:55:54,429][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:55:54,929][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:55:55,429][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:55:55,928][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:55:56,427][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:55:56,926][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:55:57,424][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:55:57,924][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:55:58,425][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:55:58,924][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:55:59,423][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:55:59,922][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:56:00,420][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:56:00,918][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:56:01,418][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:56:01,919][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:56:02,419][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:56:02,919][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:56:03,419][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:56:03,919][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:56:04,419][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:56:04,918][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:56:05,419][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:56:05,918][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:56:06,418][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:56:06,917][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:56:07,417][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:56:07,916][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:56:08,415][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:56:08,915][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:56:09,415][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:56:09,915][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:56:10,414][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:56:10,914][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:56:11,414][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:56:11,913][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:56:12,413][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:56:12,913][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:56:13,414][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:56:13,914][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:56:14,413][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:56:14,913][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:56:15,412][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:56:15,913][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:56:16,413][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:56:16,914][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:56:17,414][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:56:17,914][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:56:18,416][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:56:18,915][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:56:19,416][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:56:19,916][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:56:20,416][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:56:20,916][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 05:56:21,552][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 05:56:22,321][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:56:22,324][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:56:22,325][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:56:23,199][__main__][INFO] - Iteration 443 took 1m 14s (9.23% Gen, 89.60% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 53h 5m 59s. Estimated total time: 62h 27m 40s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 55s, 500 more iterations: 10h 24m 36s. [2026-03-26 05:56:23,201][__main__][INFO] - Starting iteration 443. [2026-03-26 05:56:23,601][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:56:23,602][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:56:29,945][__main__][INFO] - Number of regex retries in iteration 443: 0 [2026-03-26 05:56:29,946][__main__][INFO] - agents played in iteration 443 are Alice, Bob [2026-03-26 05:56:30,834][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:56:31,159][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:56:31,160][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:56:31,160][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:56:31,768][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:56:32,226][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:56:32,729][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:56:33,229][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:56:33,729][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:56:34,229][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:56:34,729][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:56:35,230][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:56:35,729][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:56:36,233][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:56:36,734][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:56:37,233][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:56:37,733][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:56:38,233][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:56:38,733][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:56:39,234][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:56:39,734][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:56:40,233][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:56:40,731][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:56:41,230][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:56:41,728][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:56:42,226][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:56:42,725][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:56:43,225][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:56:43,725][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:56:44,224][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:56:44,723][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:56:45,222][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:56:45,723][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:56:46,223][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:56:46,723][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:56:47,221][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:56:47,720][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:56:48,222][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:56:48,723][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:56:49,222][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:56:49,723][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:56:50,223][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:56:50,724][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:56:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:56:51,725][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:56:52,225][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:56:52,724][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:56:53,226][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:56:53,726][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:56:54,227][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:56:54,726][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:56:55,227][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:56:55,727][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:56:56,224][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:56:56,722][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:56:57,220][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:56:57,718][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:56:58,217][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:56:58,716][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:56:59,214][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:56:59,713][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:57:00,213][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:57:00,713][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:57:01,214][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:57:01,714][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:57:02,214][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:57:02,714][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:57:03,215][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:57:03,715][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:57:04,215][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:57:04,715][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:57:05,215][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:57:05,715][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:57:06,214][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:57:06,714][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:57:07,214][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:57:07,715][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:57:08,215][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:57:08,714][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:57:09,214][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:57:09,713][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:57:10,212][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:57:10,711][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:57:11,210][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:57:11,711][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:57:12,210][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:57:12,708][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:57:13,208][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:57:13,706][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:57:14,205][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:57:14,702][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:57:15,200][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:57:15,699][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:57:16,197][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:57:16,696][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:57:17,195][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:57:17,693][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:57:18,194][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:57:18,694][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:57:19,194][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:57:19,692][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:57:20,193][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:57:20,693][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:57:21,194][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:57:21,693][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:57:22,193][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:57:22,693][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:57:23,194][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:57:23,694][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:57:24,195][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:57:24,694][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:57:25,195][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:57:25,695][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:57:26,195][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:57:26,695][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:57:27,195][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:57:27,696][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:57:28,194][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:57:28,691][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:57:29,190][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:57:29,688][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:57:30,189][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:57:30,687][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:57:31,186][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:57:31,685][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:57:32,184][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:57:32,684][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:57:33,184][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:57:33,684][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:57:34,184][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:57:34,685][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:57:35,185][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:57:35,684][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 05:57:36,317][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 05:57:37,053][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:57:37,055][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:57:37,057][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:57:37,743][__main__][INFO] - Iteration 444 took 1m 14s (8.56% Gen, 90.51% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 52h 24m 14s. Estimated total time: 61h 47m 9s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 34s, 500 more iterations: 10h 17m 51s. [2026-03-26 05:57:37,745][__main__][INFO] - Starting iteration 444. [2026-03-26 05:57:38,148][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:57:38,148][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:57:45,026][__main__][INFO] - Number of regex retries in iteration 444: 0 [2026-03-26 05:57:45,027][__main__][INFO] - agents played in iteration 444 are Alice, Bob [2026-03-26 05:57:45,922][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:57:45,967][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:57:45,968][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:57:45,968][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:57:46,567][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:57:47,022][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:57:47,526][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:57:48,025][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:57:48,528][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:57:49,028][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:57:49,529][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:57:50,028][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:57:50,527][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:57:51,027][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:57:51,526][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:57:52,027][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:57:52,528][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:57:53,028][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:57:53,528][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:57:54,028][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:57:54,527][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:57:55,027][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:57:55,528][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:57:56,029][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:57:56,528][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:57:57,028][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:57:57,528][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:57:58,027][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:57:58,528][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:57:59,031][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:57:59,532][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:58:00,033][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:58:00,534][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:58:01,034][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:58:01,534][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:58:02,034][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:58:02,535][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:58:03,034][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:58:03,533][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:58:04,031][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:58:04,530][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:58:05,028][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:58:05,525][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:58:06,024][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:58:06,523][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:58:07,023][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:58:07,523][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:58:08,023][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:58:08,523][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:58:09,022][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:58:09,522][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:58:10,022][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:58:10,522][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:58:11,022][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:58:11,520][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:58:12,019][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:58:12,518][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:58:13,017][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:58:13,517][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:58:14,016][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:58:14,516][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:58:15,015][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:58:15,517][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:58:16,018][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:58:16,517][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:58:17,016][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:58:17,517][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:58:18,018][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:58:18,518][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:58:19,017][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:58:19,516][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:58:20,015][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:58:20,515][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:58:21,015][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:58:21,514][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:58:22,014][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:58:22,513][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:58:23,012][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:58:23,512][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:58:24,011][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:58:24,512][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:58:25,011][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:58:25,511][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:58:26,011][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:58:26,510][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:58:27,011][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:58:27,511][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:58:28,011][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:58:28,511][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:58:29,011][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:58:29,510][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:58:30,010][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:58:30,511][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:58:31,011][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:58:31,512][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:58:32,012][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:58:32,513][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:58:33,013][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:58:33,514][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:58:34,015][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:58:34,517][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:58:35,016][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:58:35,516][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:58:36,015][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:58:36,513][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:58:37,011][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:58:37,510][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:58:38,009][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:58:38,508][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:58:39,008][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:58:39,508][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:58:40,009][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:58:40,509][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:58:41,009][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:58:41,509][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:58:42,009][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:58:42,509][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:58:43,009][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:58:43,508][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 05:58:44,007][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 05:58:44,506][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 05:58:45,004][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 05:58:45,503][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 05:58:46,003][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 05:58:46,501][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 05:58:47,001][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 05:58:47,500][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 05:58:48,001][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 05:58:48,502][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 05:58:49,002][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 05:58:49,502][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 05:58:50,002][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 05:58:50,502][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21710 tokens. [2026-03-26 05:58:51,131][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 05:58:51,902][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 05:58:51,904][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 05:58:51,906][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 05:58:52,718][__main__][INFO] - Iteration 445 took 1m 14s (9.22% Gen, 89.68% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 44m 22s. Estimated total time: 62h 8m 32s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 17s, 500 more iterations: 10h 21m 25s. [2026-03-26 05:58:52,768][__main__][INFO] - Starting iteration 445. [2026-03-26 05:58:53,172][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 05:58:53,172][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 05:58:54,020][mllm.models.large_language_model_local][WARNING] - Response Proposed Allocation: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 05:59:01,109][__main__][INFO] - Number of regex retries in iteration 445: 1 [2026-03-26 05:59:01,110][__main__][INFO] - agents played in iteration 445 are Alice, Bob [2026-03-26 05:59:02,014][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:59:02,058][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 05:59:02,059][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 05:59:02,060][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 05:59:02,661][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 05:59:03,118][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 05:59:03,622][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 05:59:04,123][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 05:59:04,623][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 05:59:05,124][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 05:59:05,624][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 05:59:06,125][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 05:59:06,626][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 05:59:07,124][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 05:59:07,623][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 05:59:08,121][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 05:59:08,619][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 05:59:09,118][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 05:59:09,617][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 05:59:10,116][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 05:59:10,614][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 05:59:11,115][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 05:59:11,617][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 05:59:12,117][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 05:59:12,617][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 05:59:13,118][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 05:59:13,618][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 05:59:14,117][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 05:59:14,616][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 05:59:15,115][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 05:59:15,614][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 05:59:16,113][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 05:59:16,612][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 05:59:17,110][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 05:59:17,608][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 05:59:18,108][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 05:59:18,609][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 05:59:19,111][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 05:59:19,610][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 05:59:20,110][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 05:59:20,610][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 05:59:21,110][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 05:59:21,610][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 05:59:22,111][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 05:59:22,611][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 05:59:23,110][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 05:59:23,609][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 05:59:24,109][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 05:59:24,607][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 05:59:25,106][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 05:59:25,604][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 05:59:26,102][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 05:59:26,599][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 05:59:27,097][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 05:59:27,597][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 05:59:28,096][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 05:59:28,595][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 05:59:29,093][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 05:59:29,593][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 05:59:30,092][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 05:59:30,592][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 05:59:31,091][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 05:59:31,591][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 05:59:32,090][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 05:59:32,591][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 05:59:33,091][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 05:59:33,590][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 05:59:34,090][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 05:59:34,590][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 05:59:35,091][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 05:59:35,591][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 05:59:36,091][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 05:59:36,592][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 05:59:37,092][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 05:59:37,593][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 05:59:38,094][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 05:59:38,595][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 05:59:39,093][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 05:59:39,592][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 05:59:40,091][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 05:59:40,589][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 05:59:41,090][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 05:59:41,589][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 05:59:42,088][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 05:59:42,586][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 05:59:43,085][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 05:59:43,584][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 05:59:44,082][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 05:59:44,580][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 05:59:45,079][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 05:59:45,581][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 05:59:46,082][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 05:59:46,582][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 05:59:47,080][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 05:59:47,579][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 05:59:48,078][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 05:59:48,578][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 05:59:49,077][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 05:59:49,576][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 05:59:50,075][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 05:59:50,573][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 05:59:51,073][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 05:59:51,573][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 05:59:52,074][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 05:59:52,573][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 05:59:53,073][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 05:59:53,572][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 05:59:54,072][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 05:59:54,573][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 05:59:55,071][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 05:59:55,571][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 05:59:56,071][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 05:59:56,570][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 05:59:57,069][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 05:59:57,567][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 05:59:58,065][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 05:59:58,563][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 05:59:59,062][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 05:59:59,562][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:00:00,061][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:00:00,560][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:00:01,060][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:00:01,558][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:00:02,058][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:00:02,557][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:00:03,057][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:00:03,557][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:00:04,057][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:00:04,557][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:00:05,057][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:00:05,557][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:00:06,057][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:00:06,557][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 06:00:07,194][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:00:07,945][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:00:07,948][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:00:07,949][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:00:08,609][__main__][INFO] - Iteration 446 took 1m 15s (10.52% Gen, 88.60% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 53h 26m 26s. Estimated total time: 62h 51m 53s. Time estimates for 10 more iterations: 12m 34s, 100 more iterations: 2h 5m 43s, 500 more iterations: 10h 28m 38s. [2026-03-26 06:00:08,611][__main__][INFO] - Starting iteration 446. [2026-03-26 06:00:09,010][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 06:00:09,010][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:00:15,623][__main__][INFO] - Number of regex retries in iteration 446: 0 [2026-03-26 06:00:15,903][__main__][INFO] - agents played in iteration 446 are Alice, Bob [2026-03-26 06:00:16,777][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:00:16,822][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:00:16,823][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:00:16,823][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:00:17,426][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:00:17,881][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:00:18,386][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:00:18,885][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:00:19,385][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:00:19,884][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:00:20,383][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:00:20,881][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:00:21,380][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:00:21,879][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:00:22,376][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:00:22,875][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:00:23,375][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:00:23,875][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:00:24,373][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:00:24,871][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:00:25,371][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:00:25,870][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:00:26,369][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:00:26,868][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:00:27,369][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:00:27,868][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:00:28,368][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:00:28,867][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:00:29,366][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:00:29,864][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:00:30,362][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:00:30,859][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:00:31,358][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:00:31,859][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:00:32,357][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:00:32,857][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:00:33,354][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:00:33,856][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:00:34,357][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:00:34,858][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:00:35,359][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:00:35,859][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:00:36,359][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:00:36,860][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:00:37,360][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:00:37,858][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:00:38,357][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:00:38,856][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:00:39,355][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:00:39,854][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:00:40,352][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:00:40,851][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:00:41,350][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:00:41,851][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:00:42,350][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:00:42,850][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:00:43,349][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:00:43,849][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:00:44,348][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:00:44,846][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:00:45,344][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:00:45,843][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:00:46,341][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:00:46,841][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:00:47,340][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:00:47,839][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:00:48,338][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:00:48,837][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:00:49,338][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:00:49,838][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:00:50,337][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:00:50,837][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:00:51,337][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:00:51,836][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:00:52,335][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:00:52,835][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:00:53,335][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:00:53,834][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:00:54,333][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:00:54,832][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:00:55,332][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:00:55,832][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:00:56,331][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:00:56,830][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:00:57,330][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:00:57,829][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:00:58,329][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:00:58,828][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:00:59,327][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:00:59,825][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:01:00,324][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:01:00,824][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:01:01,323][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:01:01,821][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:01:02,320][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:01:02,818][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:01:03,316][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:01:03,815][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:01:04,313][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:01:04,812][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:01:05,310][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:01:05,810][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:01:06,311][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:01:06,813][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:01:07,314][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:01:07,815][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:01:08,316][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:01:08,818][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:01:09,318][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:01:09,818][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:01:10,317][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:01:10,816][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:01:11,315][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:01:11,814][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:01:12,313][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:01:12,812][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:01:13,311][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:01:13,809][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:01:14,310][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:01:14,810][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:01:15,310][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:01:15,813][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:01:16,313][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:01:16,812][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:01:17,311][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:01:17,811][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:01:18,313][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:01:18,813][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:01:19,312][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:01:19,812][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:01:20,311][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:01:20,810][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:01:21,311][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 06:01:21,939][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:01:22,681][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:01:22,684][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:01:22,685][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:01:23,390][__main__][INFO] - Iteration 447 took 1m 14s (9.27% Gen, 89.78% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 32m 23s. Estimated total time: 61h 59m 4s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 58s, 500 more iterations: 10h 19m 50s. [2026-03-26 06:01:23,393][__main__][INFO] - Starting iteration 447. [2026-03-26 06:01:23,797][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 06:01:23,798][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:01:30,879][__main__][INFO] - Number of regex retries in iteration 447: 0 [2026-03-26 06:01:30,880][__main__][INFO] - agents played in iteration 447 are Alice, Bob [2026-03-26 06:01:31,758][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:01:31,803][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:01:31,803][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:01:31,804][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:01:32,411][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:01:32,869][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:01:33,373][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:01:33,874][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:01:34,374][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:01:34,875][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:01:35,375][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:01:35,874][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:01:36,375][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:01:36,876][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:01:37,375][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:01:37,876][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:01:38,375][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:01:38,874][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:01:39,374][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:01:39,873][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:01:40,372][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:01:40,872][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:01:41,372][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:01:41,872][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:01:42,372][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:01:42,872][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:01:43,372][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:01:43,873][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:01:44,373][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:01:44,872][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:01:45,371][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:01:45,871][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:01:46,370][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:01:46,870][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:01:47,371][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:01:47,871][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:01:48,373][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:01:48,873][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:01:49,373][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:01:49,873][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:01:50,373][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:01:50,873][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:01:51,372][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:01:51,872][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:01:52,371][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:01:52,871][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:01:53,371][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:01:53,871][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:01:54,371][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:01:54,871][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:01:55,371][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:01:55,871][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:01:56,372][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:01:56,872][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:01:57,372][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:01:57,872][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:01:58,372][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:01:58,872][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:01:59,371][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:01:59,869][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:02:00,368][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:02:00,871][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:02:01,371][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:02:01,872][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:02:02,373][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:02:02,873][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:02:03,373][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:02:03,873][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:02:04,374][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:02:04,875][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:02:05,376][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:02:05,876][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:02:06,376][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:02:06,877][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:02:07,377][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:02:07,877][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:02:08,378][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:02:08,878][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:02:09,376][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:02:09,875][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:02:10,374][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:02:10,873][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:02:11,372][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:02:11,871][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:02:12,371][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:02:12,871][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:02:13,370][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:02:13,870][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:02:14,370][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:02:14,870][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:02:15,371][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:02:15,872][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:02:16,372][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:02:16,872][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:02:17,372][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:02:17,871][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:02:18,374][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:02:18,873][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:02:19,372][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:02:19,871][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:02:20,370][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:02:20,870][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:02:21,370][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:02:21,869][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:02:22,370][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:02:22,870][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:02:23,371][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:02:23,870][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:02:24,370][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:02:24,869][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:02:25,369][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:02:25,871][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:02:26,371][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:02:26,871][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:02:27,370][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:02:27,870][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:02:28,370][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:02:28,869][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:02:29,370][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:02:29,870][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:02:30,370][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:02:30,869][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:02:31,368][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:02:31,867][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:02:32,366][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:02:32,867][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:02:33,368][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:02:33,869][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:02:34,370][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:02:34,870][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:02:35,371][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:02:35,872][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:02:36,373][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 06:02:37,003][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 06:02:37,740][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:02:37,742][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:02:37,744][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:02:38,398][__main__][INFO] - Iteration 448 took 1m 14s (9.49% Gen, 89.63% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 42m 9s. Estimated total time: 62h 10m 5s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 20s, 500 more iterations: 10h 21m 40s. [2026-03-26 06:02:38,400][__main__][INFO] - Starting iteration 448. [2026-03-26 06:02:38,802][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 06:02:38,803][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:02:40,521][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:02:41,651][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:02:45,688][__main__][INFO] - Number of regex retries in iteration 448: 2 [2026-03-26 06:02:45,689][__main__][INFO] - agents played in iteration 448 are Alice, Bob [2026-03-26 06:02:46,568][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:02:46,613][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:02:46,614][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:02:46,614][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:02:47,217][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:02:47,673][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:02:48,179][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:02:48,680][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:02:49,180][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:02:49,680][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:02:50,180][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:02:50,680][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:02:51,180][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:02:51,679][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:02:52,178][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:02:52,677][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:02:53,177][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:02:53,677][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:02:54,176][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:02:54,676][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:02:55,176][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:02:55,677][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:02:56,176][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:02:56,676][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:02:57,176][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:02:57,677][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:02:58,177][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:02:58,676][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:02:59,175][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:02:59,675][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:03:00,176][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:03:00,675][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:03:01,175][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:03:01,674][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:03:02,174][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:03:02,673][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:03:03,172][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:03:03,672][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:03:04,171][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:03:04,670][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:03:05,170][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:03:05,670][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:03:06,168][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:03:06,667][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:03:07,168][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:03:07,668][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:03:08,167][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:03:08,666][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:03:09,165][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:03:09,665][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:03:10,163][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:03:10,661][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:03:11,161][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:03:11,658][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:03:12,157][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:03:12,655][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:03:13,153][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:03:13,651][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:03:14,149][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:03:14,646][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:03:15,144][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:03:15,644][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:03:16,143][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:03:16,642][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:03:17,141][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:03:17,640][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:03:18,140][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:03:18,642][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:03:19,141][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:03:19,641][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:03:20,142][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:03:20,642][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:03:21,143][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:03:21,643][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:03:22,142][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:03:22,642][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:03:23,142][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:03:23,642][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:03:24,141][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:03:24,639][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:03:25,139][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:03:25,638][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:03:26,138][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:03:26,637][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:03:27,137][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:03:27,637][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:03:28,137][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:03:28,636][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:03:29,136][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:03:29,635][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:03:30,135][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:03:30,635][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:03:31,136][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:03:31,636][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:03:32,134][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:03:32,634][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:03:33,133][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:03:33,632][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:03:34,131][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:03:34,631][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:03:35,130][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:03:35,629][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:03:36,128][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:03:36,627][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:03:37,127][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:03:37,626][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:03:38,125][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:03:38,625][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:03:39,123][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:03:39,622][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:03:40,120][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:03:40,618][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:03:41,118][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:03:41,617][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:03:42,117][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:03:42,616][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:03:43,115][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:03:43,613][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:03:44,111][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:03:44,610][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:03:45,109][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:03:45,609][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:03:46,107][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:03:46,604][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:03:47,102][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:03:47,600][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:03:48,099][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:03:48,599][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:03:49,098][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:03:49,598][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:03:50,097][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:03:50,595][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:03:51,096][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 06:03:51,731][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 06:03:52,492][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:03:52,494][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:03:52,496][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:03:53,327][__main__][INFO] - Iteration 449 took 1m 14s (9.24% Gen, 89.64% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 37m 6s. Estimated total time: 62h 6m 17s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 12s, 500 more iterations: 10h 21m 2s. [2026-03-26 06:03:53,330][__main__][INFO] - Starting iteration 449. [2026-03-26 06:03:53,732][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 06:03:53,732][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:04:00,622][__main__][INFO] - Number of regex retries in iteration 449: 0 [2026-03-26 06:04:00,623][__main__][INFO] - agents played in iteration 449 are Alice, Bob [2026-03-26 06:04:01,502][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:04:01,546][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:04:01,547][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:04:01,548][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:04:02,171][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:04:02,629][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:04:03,134][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:04:03,635][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:04:04,136][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:04:04,636][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:04:05,136][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:04:05,637][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:04:06,137][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:04:06,638][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:04:07,138][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:04:07,639][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:04:08,139][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:04:08,640][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:04:09,140][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:04:09,640][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:04:10,140][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:04:10,640][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:04:11,140][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:04:11,640][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:04:12,140][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:04:12,640][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:04:13,139][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:04:13,639][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:04:14,139][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:04:14,638][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:04:15,137][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:04:15,639][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:04:16,140][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:04:16,639][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:04:17,138][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:04:17,637][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:04:18,137][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:04:18,638][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:04:19,138][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:04:19,636][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:04:20,136][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:04:20,635][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:04:21,133][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:04:21,634][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:04:22,133][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:04:22,634][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:04:23,134][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:04:23,635][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:04:24,135][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:04:24,634][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:04:25,134][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:04:25,634][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:04:26,134][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:04:26,633][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:04:27,132][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:04:27,631][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:04:28,130][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:04:28,628][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:04:29,127][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:04:29,627][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:04:30,126][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:04:30,625][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:04:31,125][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:04:31,624][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:04:32,123][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:04:32,621][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:04:33,120][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:04:33,619][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:04:34,120][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:04:34,620][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:04:35,121][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:04:35,620][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:04:36,119][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:04:36,619][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:04:37,120][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:04:37,621][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:04:38,121][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:04:38,621][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:04:39,121][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:04:39,619][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:04:40,119][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:04:40,620][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:04:41,121][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:04:41,621][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:04:42,120][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:04:42,619][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:04:43,119][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:04:43,618][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:04:44,118][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:04:44,618][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:04:45,117][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:04:45,617][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:04:46,117][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:04:46,616][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:04:47,115][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:04:47,614][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:04:48,114][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:04:48,615][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:04:49,114][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:04:49,613][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:04:50,111][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:04:50,611][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:04:51,111][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:04:51,613][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:04:52,112][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:04:52,612][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:04:53,112][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:04:53,611][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:04:54,110][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:04:54,610][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:04:55,113][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:04:55,614][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:04:56,113][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:04:56,613][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:04:57,113][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:04:57,613][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:04:58,113][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:04:58,614][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:04:59,115][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:04:59,615][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:05:00,115][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:05:00,615][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:05:01,116][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:05:01,616][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:05:02,116][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:05:02,616][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:05:03,116][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:05:03,614][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:05:04,114][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:05:04,614][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:05:05,113][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:05:05,614][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:05:06,114][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 06:05:06,759][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 06:05:07,518][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:05:07,521][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:05:07,522][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:05:08,344][__main__][INFO] - Iteration 450 took 1m 14s (9.24% Gen, 89.66% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 40m 13s. Estimated total time: 62h 10m 40s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 21s, 500 more iterations: 10h 21m 46s. [2026-03-26 06:05:08,347][__main__][INFO] - Starting iteration 450. [2026-03-26 06:05:08,748][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 8 and human policies 1. [2026-03-26 06:05:08,748][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:05:15,507][__main__][INFO] - Number of regex retries in iteration 450: 0 [2026-03-26 06:05:15,508][__main__][INFO] - agents played in iteration 450 are Alice, Bob [2026-03-26 06:05:16,382][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:05:16,427][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:05:16,428][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:05:16,429][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:05:17,039][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:05:17,496][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:05:18,004][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:05:18,505][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:05:19,004][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:05:19,503][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:05:20,003][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:05:20,502][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:05:21,001][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:05:21,503][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:05:22,003][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:05:22,502][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:05:23,002][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:05:23,501][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:05:24,000][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:05:24,500][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:05:25,001][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:05:25,500][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:05:25,999][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:05:26,498][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:05:26,997][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:05:27,496][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:05:27,995][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:05:28,494][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:05:28,993][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:05:29,493][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:05:29,992][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:05:30,492][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:05:30,992][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:05:31,491][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:05:31,991][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:05:32,493][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:05:32,992][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:05:33,494][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:05:33,996][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:05:34,496][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:05:34,997][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:05:35,498][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:05:35,999][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:05:36,501][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:05:37,001][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:05:37,500][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:05:37,999][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:05:38,497][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:05:38,996][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:05:39,496][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:05:39,995][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:05:40,494][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:05:40,993][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:05:41,494][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:05:41,995][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:05:42,495][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:05:42,995][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:05:43,495][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:05:43,997][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:05:44,497][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:05:44,998][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:05:45,501][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:05:46,004][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:05:46,503][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:05:47,003][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:05:47,503][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:05:48,004][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:05:48,504][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:05:49,003][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:05:49,501][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:05:50,000][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:05:50,497][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:05:50,995][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:05:51,492][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:05:51,990][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:05:52,487][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:05:52,984][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:05:53,482][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:05:53,982][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:05:54,481][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:05:54,980][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:05:55,479][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:05:55,978][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:05:56,478][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:05:56,976][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:05:57,475][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:05:57,974][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:05:58,474][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:05:58,973][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:05:59,470][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:05:59,969][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:06:00,468][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:06:00,967][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:06:01,466][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:06:01,966][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:06:02,465][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:06:02,965][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:06:03,464][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:06:03,965][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:06:04,464][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:06:04,963][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:06:05,463][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:06:05,963][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:06:06,464][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:06:06,964][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:06:07,465][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:06:07,965][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:06:08,466][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:06:08,966][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:06:09,465][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:06:09,964][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:06:10,463][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:06:10,964][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:06:11,463][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:06:11,962][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:06:12,461][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:06:12,960][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:06:13,460][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:06:13,960][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:06:14,461][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:06:14,962][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:06:15,462][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:06:15,962][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:06:16,463][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:06:16,964][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:06:17,463][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:06:17,965][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:06:18,467][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:06:18,966][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:06:19,466][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:06:19,966][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:06:20,466][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:06:20,966][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 06:06:21,604][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:06:22,342][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:06:22,345][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:06:22,346][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:06:23,664][__main__][INFO] - Iteration 451 took 1m 14s (9.02% Gen, 89.22% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 54m 8s. Estimated total time: 62h 25m 50s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 51s, 500 more iterations: 10h 24m 18s. [2026-03-26 06:06:23,667][__main__][INFO] - Starting iteration 451. [2026-03-26 06:06:24,070][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:06:24,071][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:06:24,629][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:06:26,215][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:06:30,562][__main__][INFO] - Number of regex retries in iteration 451: 2 [2026-03-26 06:06:30,563][__main__][INFO] - agents played in iteration 451 are Alice, Bob [2026-03-26 06:06:31,445][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:06:31,490][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:06:31,490][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:06:31,491][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:06:32,094][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:06:32,549][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:06:33,053][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:06:33,553][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:06:34,052][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:06:34,552][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:06:35,050][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:06:35,549][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:06:36,049][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:06:36,550][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:06:37,050][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:06:37,549][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:06:38,049][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:06:38,549][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:06:39,047][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:06:39,546][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:06:40,045][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:06:40,566][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:06:41,067][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:06:41,567][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:06:42,067][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:06:42,566][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:06:43,066][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:06:43,566][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:06:44,066][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:06:44,565][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:06:45,063][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:06:45,562][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:06:46,060][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:06:46,559][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:06:47,057][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:06:47,557][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:06:48,057][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:06:48,558][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:06:49,057][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:06:49,557][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:06:50,056][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:06:50,555][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:06:51,054][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:06:51,554][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:06:52,053][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:06:52,553][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:06:53,053][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:06:53,552][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:06:54,052][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:06:54,551][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:06:55,052][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:06:55,551][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:06:56,050][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:06:56,548][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:06:57,047][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:06:57,546][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:06:58,045][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:06:58,544][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:06:59,043][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:06:59,543][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:07:00,042][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:07:00,543][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:07:01,043][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:07:01,544][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:07:02,046][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:07:02,547][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:07:03,047][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:07:03,547][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:07:04,048][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:07:04,547][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:07:05,046][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:07:05,545][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:07:06,044][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:07:06,543][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:07:07,043][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:07:07,542][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:07:08,041][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:07:08,539][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:07:09,039][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:07:09,538][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:07:10,038][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:07:10,537][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:07:11,036][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:07:11,534][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:07:12,034][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:07:12,534][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:07:13,035][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:07:13,536][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:07:14,036][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:07:14,535][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:07:15,034][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:07:15,535][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:07:16,035][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:07:16,537][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:07:17,040][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:07:17,540][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:07:18,040][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:07:18,541][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:07:19,041][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:07:19,541][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:07:20,040][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:07:20,539][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:07:21,038][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:07:21,538][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:07:22,037][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:07:22,536][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:07:23,035][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:07:23,535][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:07:24,034][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:07:24,533][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:07:25,032][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:07:25,531][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:07:26,031][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:07:26,530][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:07:27,030][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:07:27,529][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:07:28,029][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:07:28,528][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:07:29,027][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:07:29,526][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:07:30,025][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:07:30,524][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:07:31,024][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:07:31,523][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:07:32,022][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:07:32,522][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:07:33,022][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:07:33,523][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:07:34,023][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:07:34,524][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:07:35,026][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:07:35,526][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:07:36,027][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 06:07:36,657][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 06:07:37,397][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:07:37,400][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:07:37,401][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:07:38,107][__main__][INFO] - Iteration 452 took 1m 14s (8.77% Gen, 90.28% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 8m 56s. Estimated total time: 61h 41m 52s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 23s, 500 more iterations: 10h 16m 58s. [2026-03-26 06:07:38,109][__main__][INFO] - Starting iteration 452. [2026-03-26 06:07:38,510][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:07:38,511][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:07:45,215][__main__][INFO] - Number of regex retries in iteration 452: 0 [2026-03-26 06:07:45,215][__main__][INFO] - agents played in iteration 452 are Alice, Bob [2026-03-26 06:07:46,092][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:07:46,137][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:07:46,138][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:07:46,139][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:07:46,753][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:07:47,209][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:07:47,715][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:07:48,216][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:07:48,716][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:07:49,216][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:07:49,715][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:07:50,214][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:07:50,714][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:07:51,214][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:07:51,713][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:07:52,212][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:07:52,712][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:07:53,210][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:07:53,709][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:07:54,207][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:07:54,705][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:07:55,206][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:07:55,706][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:07:56,205][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:07:56,704][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:07:57,204][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:07:57,702][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:07:58,201][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:07:58,701][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:07:59,200][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:07:59,699][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:08:00,199][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:08:00,698][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:08:01,198][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:08:01,697][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:08:02,196][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:08:02,696][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:08:03,196][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:08:03,695][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:08:04,195][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:08:04,695][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:08:05,193][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:08:05,691][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:08:06,190][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:08:06,689][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:08:07,185][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:08:07,682][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:08:08,180][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:08:08,678][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:08:09,176][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:08:09,674][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:08:10,171][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:08:10,669][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:08:11,170][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:08:11,670][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:08:12,170][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:08:12,670][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:08:13,170][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:08:13,669][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:08:14,168][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:08:14,668][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:08:15,166][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:08:15,664][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:08:16,163][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:08:16,662][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:08:17,161][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:08:17,660][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:08:18,159][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:08:18,660][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:08:19,160][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:08:19,660][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:08:20,159][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:08:20,659][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:08:21,158][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:08:21,657][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:08:22,157][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:08:22,656][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:08:23,155][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:08:23,654][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:08:24,153][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:08:24,651][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:08:25,150][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:08:25,648][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:08:26,146][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:08:26,644][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:08:27,144][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:08:27,643][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:08:28,142][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:08:28,642][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:08:29,141][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:08:29,640][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:08:30,140][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:08:30,640][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:08:31,140][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:08:31,638][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:08:32,138][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:08:32,639][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:08:33,138][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:08:33,638][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:08:34,138][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:08:34,638][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:08:35,137][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:08:35,635][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:08:36,132][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:08:36,630][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:08:37,129][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:08:37,627][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:08:38,125][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:08:38,623][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:08:39,120][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:08:39,617][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:08:40,114][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:08:40,612][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:08:41,109][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:08:41,605][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:08:42,101][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:08:42,598][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:08:43,099][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:08:43,599][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:08:44,100][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:08:44,599][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:08:45,098][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:08:45,601][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:08:46,101][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:08:46,601][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:08:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:08:47,600][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:08:48,099][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:08:48,599][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:08:49,098][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:08:49,597][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:08:50,096][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:08:50,595][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:08:51,231][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 06:08:51,986][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:08:51,989][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:08:51,990][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:08:52,654][__main__][INFO] - Iteration 453 took 1m 14s (9.04% Gen, 90.06% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 13m 5s. Estimated total time: 61h 47m 15s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 34s, 500 more iterations: 10h 17m 52s. [2026-03-26 06:08:52,656][__main__][INFO] - Starting iteration 453. [2026-03-26 06:08:53,057][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:08:53,058][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:08:58,469][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:08:59,733][__main__][INFO] - Number of regex retries in iteration 453: 1 [2026-03-26 06:08:59,734][__main__][INFO] - agents played in iteration 453 are Alice, Bob [2026-03-26 06:09:00,889][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:09:00,934][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:09:00,935][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:09:00,935][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:09:01,539][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:09:01,993][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:09:02,497][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:09:02,995][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:09:03,495][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:09:03,993][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:09:04,492][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:09:04,991][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:09:05,490][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:09:05,990][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:09:06,491][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:09:06,990][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:09:07,490][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:09:07,990][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:09:08,488][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:09:08,987][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:09:09,487][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:09:09,988][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:09:10,487][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:09:10,986][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:09:11,484][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:09:11,984][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:09:12,482][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:09:12,981][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:09:13,482][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:09:14,000][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:09:14,505][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:09:15,004][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:09:15,504][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:09:16,004][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:09:16,503][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:09:17,004][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:09:17,504][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:09:18,005][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:09:18,506][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:09:19,007][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:09:19,507][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:09:20,007][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:09:20,508][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:09:21,009][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:09:21,509][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:09:22,007][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:09:22,506][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:09:23,005][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:09:23,504][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:09:24,003][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:09:24,501][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:09:25,000][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:09:25,500][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:09:26,000][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:09:26,500][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:09:27,000][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:09:27,500][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:09:27,999][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:09:28,499][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:09:28,998][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:09:29,497][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:09:29,995][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:09:30,494][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:09:30,993][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:09:31,492][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:09:31,991][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:09:32,489][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:09:32,987][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:09:33,485][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:09:33,983][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:09:34,481][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:09:34,980][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:09:35,478][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:09:35,976][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:09:36,475][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:09:36,974][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:09:37,472][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:09:37,971][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:09:38,471][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:09:38,971][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:09:39,471][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:09:39,970][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:09:40,468][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:09:40,967][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:09:41,466][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:09:41,966][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:09:42,465][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:09:42,964][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:09:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:09:43,963][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:09:44,463][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:09:44,962][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:09:45,462][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:09:45,962][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:09:46,462][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:09:46,962][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:09:47,462][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:09:47,963][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:09:48,465][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:09:48,965][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:09:49,464][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:09:49,964][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:09:50,463][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:09:50,963][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:09:51,462][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:09:51,962][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:09:52,461][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:09:52,961][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:09:53,462][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:09:53,961][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:09:54,460][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:09:54,959][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:09:55,457][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:09:55,957][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:09:56,458][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:09:56,957][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:09:57,456][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:09:57,956][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:09:58,456][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:09:58,956][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:09:59,456][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:09:59,956][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:10:00,456][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:10:00,956][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:10:01,456][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:10:01,954][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:10:02,452][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:10:02,950][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:10:03,447][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:10:03,946][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:10:04,443][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:10:04,942][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:10:05,443][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 06:10:06,090][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 06:10:06,826][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:10:06,828][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:10:06,830][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:10:07,481][__main__][INFO] - Iteration 454 took 1m 14s (9.34% Gen, 89.78% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 25m 49s. Estimated total time: 62h 1m 14s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 2s, 500 more iterations: 10h 20m 12s. [2026-03-26 06:10:07,484][__main__][INFO] - Starting iteration 454. [2026-03-26 06:10:07,887][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:10:07,887][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:10:08,477][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:10:14,929][__main__][INFO] - Number of regex retries in iteration 454: 1 [2026-03-26 06:10:14,930][__main__][INFO] - agents played in iteration 454 are Alice, Bob [2026-03-26 06:10:15,829][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:10:15,874][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:10:15,874][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:10:15,875][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:10:16,485][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:10:16,938][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:10:17,441][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:10:17,941][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:10:18,443][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:10:18,941][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:10:19,440][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:10:19,938][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:10:20,437][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:10:20,938][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:10:21,437][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:10:21,936][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:10:22,437][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:10:22,936][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:10:23,435][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:10:23,935][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:10:24,434][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:10:24,934][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:10:25,434][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:10:25,934][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:10:26,434][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:10:26,935][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:10:27,434][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:10:27,933][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:10:28,432][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:10:28,931][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:10:29,431][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:10:29,930][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:10:30,429][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:10:30,928][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:10:31,426][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:10:31,924][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:10:32,423][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:10:32,921][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:10:33,420][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:10:33,919][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:10:34,419][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:10:34,918][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:10:35,416][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:10:35,915][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:10:36,414][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:10:36,932][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:10:37,437][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:10:37,937][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:10:38,437][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:10:38,937][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:10:39,437][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:10:39,937][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:10:40,437][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:10:40,936][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:10:41,434][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:10:41,932][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:10:42,431][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:10:42,929][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:10:43,428][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:10:43,927][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:10:44,426][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:10:44,925][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:10:45,422][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:10:45,920][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:10:46,419][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:10:46,918][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:10:47,415][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:10:47,915][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:10:48,416][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:10:48,915][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:10:49,414][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:10:49,912][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:10:50,411][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:10:50,909][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:10:51,407][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:10:51,906][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:10:52,405][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:10:52,906][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:10:53,405][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:10:53,904][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:10:54,403][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:10:54,901][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:10:55,401][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:10:55,901][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:10:56,400][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:10:56,900][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:10:57,399][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:10:57,899][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:10:58,399][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:10:58,897][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:10:59,398][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:10:59,897][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:11:00,397][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:11:00,895][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:11:01,394][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:11:01,893][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:11:02,392][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:11:02,891][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:11:03,391][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:11:03,890][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:11:04,389][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:11:04,888][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:11:05,386][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:11:05,885][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:11:06,383][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:11:06,882][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:11:07,381][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:11:07,880][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:11:08,379][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:11:08,878][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:11:09,378][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:11:09,878][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:11:10,379][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:11:10,878][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:11:11,379][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:11:11,879][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:11:12,378][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:11:12,876][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:11:13,375][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:11:13,872][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:11:14,371][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:11:14,869][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:11:15,367][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:11:15,865][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:11:16,363][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:11:16,861][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:11:17,358][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:11:17,857][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:11:18,357][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:11:18,857][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:11:19,356][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:11:19,854][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:11:20,352][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21732 tokens. [2026-03-26 06:11:20,986][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 06:11:21,725][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:11:21,728][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:11:21,729][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:11:22,534][__main__][INFO] - Iteration 455 took 1m 14s (9.43% Gen, 89.49% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 35m 44s. Estimated total time: 62h 12m 24s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 4s. [2026-03-26 06:11:22,536][__main__][INFO] - Starting iteration 455. [2026-03-26 06:11:22,936][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:11:22,936][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:11:27,358][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:11:30,217][__main__][INFO] - Number of regex retries in iteration 455: 1 [2026-03-26 06:11:30,218][__main__][INFO] - agents played in iteration 455 are Alice, Bob [2026-03-26 06:11:31,097][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:11:31,142][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:11:31,143][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:11:31,143][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:11:31,754][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:11:32,209][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:11:32,715][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:11:33,216][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:11:33,716][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:11:34,216][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:11:34,715][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:11:35,216][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:11:35,716][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:11:36,218][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:11:36,718][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:11:37,219][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:11:37,718][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:11:38,218][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:11:38,718][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:11:39,219][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:11:39,720][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:11:40,221][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:11:40,720][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:11:41,220][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:11:41,720][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:11:42,219][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:11:42,719][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:11:43,220][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:11:43,720][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:11:44,220][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:11:44,719][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:11:45,219][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:11:45,721][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:11:46,221][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:11:46,721][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:11:47,220][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:11:47,719][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:11:48,221][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:11:48,722][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:11:49,222][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:11:49,723][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:11:50,223][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:11:50,723][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:11:51,223][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:11:51,723][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:11:52,222][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:11:52,721][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:11:53,221][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:11:53,721][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:11:54,221][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:11:54,721][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:11:55,221][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:11:55,720][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:11:56,219][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:11:56,720][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:11:57,221][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:11:57,720][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:11:58,219][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:11:58,719][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:11:59,219][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:11:59,719][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:12:00,218][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:12:00,716][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:12:01,214][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:12:01,713][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:12:02,213][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:12:02,711][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:12:03,208][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:12:03,706][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:12:04,205][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:12:04,704][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:12:05,203][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:12:05,703][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:12:06,202][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:12:06,701][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:12:07,201][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:12:07,701][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:12:08,201][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:12:08,701][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:12:09,201][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:12:09,701][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:12:10,200][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:12:10,699][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:12:11,199][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:12:11,698][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:12:12,197][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:12:12,697][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:12:13,195][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:12:13,694][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:12:14,192][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:12:14,691][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:12:15,189][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:12:15,688][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:12:16,187][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:12:16,685][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:12:17,184][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:12:17,682][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:12:18,182][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:12:18,682][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:12:19,180][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:12:19,680][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:12:20,179][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:12:20,680][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:12:21,180][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:12:21,679][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:12:22,179][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:12:22,679][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:12:23,180][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:12:23,679][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:12:24,179][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:12:24,678][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:12:25,178][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:12:25,677][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:12:26,176][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:12:26,676][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:12:27,175][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:12:27,675][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:12:28,174][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:12:28,673][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:12:29,172][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:12:29,670][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:12:30,169][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:12:30,668][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:12:31,166][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:12:31,665][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:12:32,165][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:12:32,664][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:12:33,163][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:12:33,662][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:12:34,161][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:12:34,661][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:12:35,160][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:12:35,660][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-26 06:12:36,299][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 06:12:37,144][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:12:37,147][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:12:37,148][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:12:37,835][__main__][INFO] - Iteration 456 took 1m 14s (9.72% Gen, 89.36% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 47m 4s. Estimated total time: 62h 25m 0s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 50s, 500 more iterations: 10h 24m 10s. [2026-03-26 06:12:37,837][__main__][INFO] - Starting iteration 456. [2026-03-26 06:12:38,240][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:12:38,240][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:12:44,743][__main__][INFO] - Number of regex retries in iteration 456: 0 [2026-03-26 06:12:44,744][__main__][INFO] - agents played in iteration 456 are Alice, Bob [2026-03-26 06:12:45,623][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:12:45,667][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:12:45,668][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:12:45,669][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:12:46,274][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:12:46,729][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:12:47,231][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:12:47,729][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:12:48,229][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:12:48,729][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:12:49,229][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:12:49,727][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:12:50,226][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:12:50,726][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:12:51,225][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:12:51,724][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:12:52,223][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:12:52,723][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:12:53,222][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:12:53,721][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:12:54,220][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:12:54,720][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:12:55,220][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:12:55,719][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:12:56,219][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:12:56,718][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:12:57,217][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:12:57,716][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:12:58,215][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:12:58,718][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:12:59,218][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:12:59,718][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:13:00,218][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:13:00,718][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:13:01,218][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:13:01,718][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:13:02,219][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:13:02,718][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:13:03,216][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:13:03,715][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:13:04,214][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:13:04,712][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:13:05,211][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:13:05,708][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:13:06,206][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:13:06,705][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:13:07,204][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:13:07,704][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:13:08,204][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:13:08,705][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:13:09,206][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:13:09,705][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:13:10,204][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:13:10,703][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:13:11,203][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:13:11,703][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:13:12,203][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:13:12,702][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:13:13,201][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:13:13,700][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:13:14,200][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:13:14,699][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:13:15,200][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:13:15,703][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:13:16,204][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:13:16,703][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:13:17,202][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:13:17,702][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:13:18,204][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:13:18,703][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:13:19,202][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:13:19,701][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:13:20,199][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:13:20,697][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:13:21,197][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:13:21,696][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:13:22,195][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:13:22,695][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:13:23,194][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:13:23,694][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:13:24,192][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:13:24,692][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:13:25,191][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:13:25,691][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:13:26,191][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:13:26,691][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:13:27,190][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:13:27,689][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:13:28,188][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:13:28,687][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:13:29,186][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:13:29,686][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:13:30,185][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:13:30,685][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:13:31,185][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:13:31,684][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:13:32,185][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:13:32,684][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:13:33,184][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:13:33,684][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:13:34,184][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:13:34,683][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:13:35,181][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:13:35,679][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:13:36,178][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:13:36,677][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:13:37,177][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:13:37,675][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:13:38,175][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:13:38,675][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:13:39,175][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:13:39,674][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:13:40,174][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:13:40,676][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:13:41,176][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:13:41,675][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:13:42,175][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:13:42,675][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:13:43,174][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:13:43,673][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:13:44,173][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:13:44,673][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:13:45,172][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:13:45,671][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:13:46,169][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:13:46,670][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:13:47,169][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:13:47,668][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:13:48,169][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:13:48,670][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:13:49,169][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:13:49,667][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:13:50,168][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 06:13:50,804][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.26%, ΔTime: 00:01:04 [2026-03-26 06:13:51,538][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:13:51,541][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:13:51,542][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:13:52,212][__main__][INFO] - Iteration 457 took 1m 13s (8.79% Gen, 90.30% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 59m 29s. Estimated total time: 61h 38m 39s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 17s, 500 more iterations: 10h 16m 26s. [2026-03-26 06:13:52,215][__main__][INFO] - Starting iteration 457. [2026-03-26 06:13:52,617][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:13:52,618][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:13:59,571][__main__][INFO] - Number of regex retries in iteration 457: 0 [2026-03-26 06:13:59,572][__main__][INFO] - agents played in iteration 457 are Alice, Bob [2026-03-26 06:14:00,454][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:14:00,499][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:14:00,499][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:14:00,500][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:14:01,099][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:14:01,558][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:14:02,063][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:14:02,561][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:14:03,060][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:14:03,560][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:14:04,060][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:14:04,560][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:14:05,060][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:14:05,559][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:14:06,057][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:14:06,556][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:14:07,056][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:14:07,557][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:14:08,057][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:14:08,556][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:14:09,055][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:14:09,554][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:14:10,054][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:14:10,551][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:14:11,049][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:14:11,547][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:14:12,046][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:14:12,544][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:14:13,043][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:14:13,541][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:14:14,040][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:14:14,538][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:14:15,037][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:14:15,535][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:14:16,034][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:14:16,533][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:14:17,031][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:14:17,529][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:14:18,029][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:14:18,530][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:14:19,028][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:14:19,526][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:14:20,025][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:14:20,524][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:14:21,022][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:14:21,523][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:14:22,019][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:14:22,519][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:14:23,018][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:14:23,518][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:14:24,017][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:14:24,517][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:14:25,017][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:14:25,517][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:14:26,018][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:14:26,518][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:14:27,018][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:14:27,518][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:14:28,019][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:14:28,520][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:14:29,027][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:14:29,525][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:14:30,023][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:14:30,522][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:14:31,020][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:14:31,517][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:14:32,015][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:14:32,513][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:14:33,010][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:14:33,508][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:14:34,006][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:14:34,505][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:14:35,003][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:14:35,502][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:14:36,001][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:14:36,500][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:14:36,998][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:14:37,497][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:14:37,996][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:14:38,498][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:14:38,997][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:14:39,496][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:14:39,995][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:14:40,493][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:14:40,991][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:14:41,489][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:14:41,987][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:14:42,485][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:14:42,984][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:14:43,482][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:14:43,980][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:14:44,478][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:14:44,976][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:14:45,476][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:14:45,977][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:14:46,477][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:14:46,976][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:14:47,474][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:14:47,974][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:14:48,475][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:14:48,974][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:14:49,473][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:14:49,972][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:14:50,472][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:14:50,971][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:14:51,469][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:14:51,967][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:14:52,465][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:14:52,964][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:14:53,463][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:14:53,962][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:14:54,461][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:14:54,961][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:14:55,461][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:14:55,960][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:14:56,459][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:14:56,961][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:14:57,460][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:14:57,960][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:14:58,460][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:14:58,959][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:14:59,458][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:14:59,957][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:15:00,457][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:15:00,956][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:15:01,455][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:15:01,954][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:15:02,452][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:15:02,949][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:15:03,446][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:15:03,944][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:15:04,441][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:15:04,941][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21723 tokens. [2026-03-26 06:15:05,584][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 06:15:06,320][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:15:06,323][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:15:06,324][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:15:07,009][__main__][INFO] - Iteration 458 took 1m 14s (9.35% Gen, 89.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 19m 11s. Estimated total time: 61h 59m 36s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 59s, 500 more iterations: 10h 19m 56s. [2026-03-26 06:15:07,011][__main__][INFO] - Starting iteration 458. [2026-03-26 06:15:07,415][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:15:07,416][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:15:13,858][__main__][INFO] - Number of regex retries in iteration 458: 0 [2026-03-26 06:15:13,859][__main__][INFO] - agents played in iteration 458 are Alice, Bob [2026-03-26 06:15:14,737][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:15:14,782][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:15:14,783][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:15:14,784][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:15:15,383][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:15:15,838][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:15:16,342][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:15:16,842][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:15:17,341][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:15:17,841][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:15:18,341][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:15:18,841][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:15:19,342][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:15:19,845][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:15:20,346][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:15:20,846][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:15:21,346][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:15:21,847][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:15:22,348][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:15:22,848][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:15:23,348][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:15:23,847][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:15:24,347][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:15:24,846][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:15:25,346][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:15:25,845][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:15:26,344][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:15:26,843][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:15:27,342][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:15:27,842][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:15:28,343][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:15:28,843][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:15:29,343][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:15:29,843][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:15:30,343][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:15:30,843][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:15:31,343][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:15:31,843][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:15:32,343][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:15:32,843][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:15:33,342][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:15:33,842][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:15:34,342][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:15:34,843][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:15:35,343][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:15:35,842][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:15:36,341][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:15:36,841][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:15:37,340][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:15:37,839][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:15:38,338][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:15:38,837][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:15:39,337][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:15:39,836][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:15:40,335][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:15:40,835][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:15:41,334][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:15:41,832][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:15:42,330][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:15:42,829][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:15:43,328][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:15:43,827][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:15:44,326][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:15:44,825][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:15:45,324][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:15:45,823][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:15:46,322][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:15:46,823][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:15:47,321][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:15:47,821][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:15:48,323][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:15:48,823][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:15:49,322][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:15:49,822][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:15:50,322][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:15:50,823][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:15:51,322][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:15:51,822][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:15:52,323][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:15:52,823][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:15:53,324][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:15:53,825][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:15:54,325][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:15:54,826][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:15:55,326][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:15:55,824][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:15:56,323][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:15:56,823][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:15:57,323][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:15:57,822][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:15:58,321][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:15:58,820][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:15:59,320][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:15:59,820][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:16:00,320][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:16:00,820][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:16:01,319][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:16:01,818][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:16:02,318][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:16:02,819][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:16:03,319][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:16:03,819][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:16:04,319][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:16:04,818][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:16:05,318][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:16:05,817][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:16:06,317][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:16:06,818][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:16:07,317][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:16:07,816][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:16:08,315][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:16:08,814][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:16:09,312][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:16:09,809][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:16:10,308][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:16:10,807][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:16:11,305][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:16:11,805][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:16:12,305][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:16:12,804][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:16:13,302][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:16:13,801][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:16:14,299][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:16:14,799][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:16:15,299][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:16:15,801][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:16:16,302][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:16:16,801][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:16:17,299][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:16:17,798][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:16:18,301][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:16:18,800][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:16:19,300][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 06:16:19,944][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:16:20,685][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:16:20,687][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:16:20,689][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:16:21,346][__main__][INFO] - Iteration 459 took 1m 13s (8.71% Gen, 90.39% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 54m 54s. Estimated total time: 61h 36m 33s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 13s, 500 more iterations: 10h 16m 5s. [2026-03-26 06:16:21,348][__main__][INFO] - Starting iteration 459. [2026-03-26 06:16:21,753][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:16:21,753][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:16:23,459][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:16:28,467][__main__][INFO] - Number of regex retries in iteration 459: 1 [2026-03-26 06:16:28,468][__main__][INFO] - agents played in iteration 459 are Alice, Bob [2026-03-26 06:16:29,345][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:16:29,389][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:16:29,390][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:16:29,391][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:16:29,997][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:16:30,453][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:16:30,957][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:16:31,457][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:16:31,957][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:16:32,458][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:16:32,958][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:16:33,458][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:16:33,958][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:16:34,457][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:16:34,957][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:16:35,457][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:16:35,958][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:16:36,458][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:16:36,958][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:16:37,458][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:16:37,957][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:16:38,456][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:16:38,955][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:16:39,454][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:16:39,953][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:16:40,452][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:16:40,950][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:16:41,448][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:16:41,946][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:16:42,446][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:16:42,945][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:16:43,444][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:16:43,943][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:16:44,443][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:16:44,942][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:16:45,441][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:16:45,939][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:16:46,439][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:16:46,939][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:16:47,439][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:16:47,940][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:16:48,442][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:16:48,943][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:16:49,443][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:16:49,943][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:16:50,443][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:16:50,943][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:16:51,443][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:16:51,943][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:16:52,443][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:16:52,942][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:16:53,444][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:16:53,943][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:16:54,446][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:16:54,946][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:16:55,447][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:16:55,947][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:16:56,447][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:16:56,947][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:16:57,447][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:16:57,947][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:16:58,451][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:16:58,952][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:16:59,452][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:16:59,952][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:17:00,452][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:17:00,954][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:17:01,454][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:17:01,956][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:17:02,456][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:17:02,957][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:17:03,456][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:17:03,957][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:17:04,457][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:17:04,957][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:17:05,457][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:17:05,956][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:17:06,456][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:17:06,956][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:17:07,457][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:17:07,959][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:17:08,460][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:17:08,960][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:17:09,460][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:17:09,961][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:17:10,460][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:17:10,959][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:17:11,459][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:17:11,959][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:17:12,460][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:17:12,960][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:17:13,460][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:17:13,960][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:17:14,461][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:17:14,962][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:17:15,463][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:17:15,963][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:17:16,465][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:17:16,967][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:17:17,468][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:17:17,970][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:17:18,473][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:17:18,974][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:17:19,474][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:17:19,974][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:17:20,478][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:17:20,980][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:17:21,482][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:17:21,984][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:17:22,485][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:17:22,987][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:17:23,488][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:17:23,989][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:17:24,496][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:17:24,999][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:17:25,502][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:17:26,005][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:17:26,508][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:17:27,011][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:17:27,513][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:17:28,014][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:17:28,523][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:17:29,028][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:17:29,531][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:17:30,034][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:17:30,537][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:17:31,039][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:17:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:17:32,045][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:17:32,554][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:17:33,057][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:17:33,560][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:17:34,063][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:17:34,701][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.66%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:17:36,882][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:17:36,885][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:17:36,886][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:17:38,464][__main__][INFO] - Iteration 460 took 1m 16s (8.75% Gen, 89.19% Train). Generation: 6s, Training: 1m 8s. Estimated remaining time: 54h 12m 40s. Estimated total time: 63h 55m 36s. Time estimates for 10 more iterations: 12m 47s, 100 more iterations: 2h 7m 51s, 500 more iterations: 10h 39m 16s. [2026-03-26 06:17:38,467][__main__][INFO] - Starting iteration 460. [2026-03-26 06:17:38,869][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:17:38,870][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:17:39,487][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:17:46,016][__main__][INFO] - Number of regex retries in iteration 460: 1 [2026-03-26 06:17:46,017][__main__][INFO] - agents played in iteration 460 are Alice, Bob [2026-03-26 06:17:46,893][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:17:46,939][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:17:46,939][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:17:46,940][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:17:47,563][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:17:48,020][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:17:48,526][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:17:49,025][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:17:49,526][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:17:50,025][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:17:50,525][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:17:51,025][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:17:51,524][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:17:52,026][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:17:52,526][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:17:53,027][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:17:53,527][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:17:54,029][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:17:54,529][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:17:55,029][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:17:55,529][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:17:56,029][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:17:56,531][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:17:57,031][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:17:57,531][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:17:58,031][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:17:58,531][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:17:59,030][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:17:59,530][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:18:00,035][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:18:00,537][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:18:01,039][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:18:01,539][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:18:02,041][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:18:02,543][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:18:03,045][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:18:03,546][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:18:04,066][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:18:04,568][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:18:05,070][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:18:05,573][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:18:06,075][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:18:06,578][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:18:07,080][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:18:07,582][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:18:08,082][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:18:08,582][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:18:09,082][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:18:09,581][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:18:10,081][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:18:10,580][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:18:11,078][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:18:11,578][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:18:12,079][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:18:12,579][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:18:13,079][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:18:13,580][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:18:14,081][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:18:14,582][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:18:15,082][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:18:15,582][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:18:16,081][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:18:16,581][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:18:17,081][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:18:17,581][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:18:18,082][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:18:18,608][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:18:19,108][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:18:19,607][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:18:20,106][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:18:20,605][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:18:21,104][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:18:21,604][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:18:22,103][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:18:22,602][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:18:23,102][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:18:23,601][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:18:24,102][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:18:24,603][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:18:25,104][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:18:25,604][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:18:26,104][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:18:26,604][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:18:27,105][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:18:27,606][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:18:28,105][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:18:28,605][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:18:29,107][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:18:29,607][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:18:30,107][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:18:30,607][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:18:31,107][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:18:31,607][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:18:32,108][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:18:32,608][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:18:33,113][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:18:33,617][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:18:34,118][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:18:34,620][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:18:35,122][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:18:35,624][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:18:36,128][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:18:36,634][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:18:37,139][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:18:37,645][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:18:38,150][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:18:38,655][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:18:39,160][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:18:39,665][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:18:40,166][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:18:40,666][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:18:41,170][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:18:41,673][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:18:42,174][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:18:42,676][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:18:43,178][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:18:43,679][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:18:44,182][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:18:44,684][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:18:45,190][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:18:45,692][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:18:46,194][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:18:46,697][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:18:47,199][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:18:47,702][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:18:48,206][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:18:48,708][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:18:49,215][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:18:49,720][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:18:50,222][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:18:50,724][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:18:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:18:51,729][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21743 tokens. [2026-03-26 06:18:52,385][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 06:18:53,941][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:18:53,943][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:18:53,945][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:18:54,969][__main__][INFO] - Iteration 461 took 1m 16s (9.39% Gen, 89.26% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 53h 40m 46s. Estimated total time: 63h 24m 59s. Time estimates for 10 more iterations: 12m 40s, 100 more iterations: 2h 6m 49s, 500 more iterations: 10h 34m 9s. [2026-03-26 06:18:54,971][__main__][INFO] - Starting iteration 461. [2026-03-26 06:18:55,379][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:18:55,380][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:18:56,203][mllm.models.large_language_model_local][WARNING] - Response Proposed Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:18:57,526][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:19:01,876][__main__][INFO] - Number of regex retries in iteration 461: 2 [2026-03-26 06:19:01,876][__main__][INFO] - agents played in iteration 461 are Alice, Bob [2026-03-26 06:19:02,756][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:19:03,087][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:19:03,088][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:19:03,088][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:19:03,699][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:19:04,153][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:19:04,657][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:19:05,156][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:19:05,656][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:19:06,155][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:19:06,654][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:19:07,154][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:19:07,655][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:19:08,155][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:19:08,655][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:19:09,154][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:19:09,653][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:19:10,151][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:19:10,651][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:19:11,149][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:19:11,647][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:19:12,147][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:19:12,646][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:19:13,144][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:19:13,643][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:19:14,143][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:19:14,642][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:19:15,141][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:19:15,640][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:19:16,142][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:19:16,646][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:19:17,148][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:19:17,648][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:19:18,149][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:19:18,650][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:19:19,151][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:19:19,651][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:19:20,151][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:19:20,650][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:19:21,150][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:19:21,649][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:19:22,148][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:19:22,647][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:19:23,147][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:19:23,647][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:19:24,145][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:19:24,644][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:19:25,144][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:19:25,644][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:19:26,143][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:19:26,643][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:19:27,143][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:19:27,643][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:19:28,142][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:19:28,641][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:19:29,139][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:19:29,638][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:19:30,137][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:19:30,636][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:19:31,137][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:19:31,635][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:19:32,133][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:19:32,633][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:19:33,132][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:19:33,631][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:19:34,131][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:19:34,630][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:19:35,129][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:19:35,628][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:19:36,127][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:19:36,625][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:19:37,125][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:19:37,625][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:19:38,125][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:19:38,624][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:19:39,123][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:19:39,622][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:19:40,122][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:19:40,621][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:19:41,120][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:19:41,619][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:19:42,117][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:19:42,616][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:19:43,115][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:19:43,615][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:19:44,114][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:19:44,613][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:19:45,111][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:19:45,610][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:19:46,109][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:19:46,608][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:19:47,106][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:19:47,605][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:19:48,105][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:19:48,607][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:19:49,107][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:19:49,606][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:19:50,105][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:19:50,605][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:19:51,104][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:19:51,605][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:19:52,105][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:19:52,606][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:19:53,105][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:19:53,605][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:19:54,103][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:19:54,602][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:19:55,103][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:19:55,602][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:19:56,103][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:19:56,603][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:19:57,102][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:19:57,602][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:19:58,102][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:19:58,602][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:19:59,101][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:19:59,601][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:20:00,106][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:20:00,607][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:20:01,107][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:20:01,608][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:20:02,108][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:20:02,607][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:20:03,108][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:20:03,609][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:20:04,109][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:20:04,613][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:20:05,114][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:20:05,615][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:20:06,116][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:20:06,619][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:20:07,121][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:20:07,623][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:20:08,272][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 06:20:09,048][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:20:09,050][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:20:09,052][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:20:09,687][__main__][INFO] - Iteration 462 took 1m 14s (8.74% Gen, 90.40% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 52h 9m 57s. Estimated total time: 61h 55m 24s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 50s, 500 more iterations: 10h 19m 14s. [2026-03-26 06:20:09,689][__main__][INFO] - Starting iteration 462. [2026-03-26 06:20:10,089][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:20:10,090][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:20:16,561][__main__][INFO] - Number of regex retries in iteration 462: 0 [2026-03-26 06:20:16,562][__main__][INFO] - agents played in iteration 462 are Alice, Bob [2026-03-26 06:20:17,439][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:20:17,484][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:20:17,484][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:20:17,485][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:20:18,103][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:20:18,559][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:20:19,064][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:20:19,564][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:20:20,063][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:20:20,563][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:20:21,063][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:20:21,563][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:20:22,063][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:20:22,563][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:20:23,063][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:20:23,562][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:20:24,062][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:20:24,561][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:20:25,062][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:20:25,562][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:20:26,063][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:20:26,562][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:20:27,060][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:20:27,557][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:20:28,055][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:20:28,552][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:20:29,050][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:20:29,548][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:20:30,046][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:20:30,546][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:20:31,045][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:20:31,545][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:20:32,044][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:20:32,544][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:20:33,043][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:20:33,543][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:20:34,043][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:20:34,542][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:20:35,040][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:20:35,538][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:20:36,036][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:20:36,533][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:20:37,032][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:20:37,530][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:20:38,029][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:20:38,531][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:20:39,031][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:20:39,532][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:20:40,033][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:20:40,533][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:20:41,034][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:20:41,536][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:20:42,038][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:20:42,537][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:20:43,037][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:20:43,538][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:20:44,039][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:20:44,539][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:20:45,040][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:20:45,543][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:20:46,045][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:20:46,543][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:20:47,042][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:20:47,541][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:20:48,040][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:20:48,540][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:20:49,039][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:20:49,537][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:20:50,034][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:20:50,532][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:20:51,032][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:20:51,531][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:20:52,031][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:20:52,531][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:20:53,031][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:20:53,530][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:20:54,029][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:20:54,529][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:20:55,030][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:20:55,530][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:20:56,030][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:20:56,530][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:20:57,030][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:20:57,529][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:20:58,035][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:20:58,535][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:20:59,035][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:20:59,535][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:21:00,035][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:21:00,534][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:21:01,032][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:21:01,531][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:21:02,033][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:21:02,534][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:21:03,034][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:21:03,535][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:21:04,035][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:21:04,535][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:21:05,037][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:21:05,538][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:21:06,043][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:21:06,544][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:21:07,044][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:21:07,545][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:21:08,045][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:21:08,545][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:21:09,045][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:21:09,545][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:21:10,049][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:21:10,551][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:21:11,053][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:21:11,556][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:21:12,059][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:21:12,562][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:21:13,064][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:21:13,567][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:21:14,074][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:21:14,577][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:21:15,080][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:21:15,582][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:21:16,084][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:21:16,587][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:21:17,089][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:21:17,591][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:21:18,100][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:21:18,603][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:21:19,105][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:21:19,605][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:21:20,106][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:21:20,606][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:21:21,106][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:21:21,606][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:21:22,106][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 06:21:22,764][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.52%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 06:21:23,514][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:21:23,517][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:21:23,518][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:21:24,247][__main__][INFO] - Iteration 463 took 1m 14s (8.73% Gen, 90.29% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 1m 12s. Estimated total time: 61h 47m 54s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 59s. [2026-03-26 06:21:24,249][__main__][INFO] - Starting iteration 463. [2026-03-26 06:21:24,653][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:21:24,654][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:21:25,293][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:21:31,732][__main__][INFO] - Number of regex retries in iteration 463: 1 [2026-03-26 06:21:31,733][__main__][INFO] - agents played in iteration 463 are Alice, Bob [2026-03-26 06:21:32,613][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:21:32,658][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:21:32,658][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:21:32,659][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:21:33,267][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:21:33,722][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:21:34,226][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:21:34,726][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:21:35,226][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:21:35,724][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:21:36,224][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:21:36,723][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:21:37,222][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:21:37,721][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:21:38,220][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:21:38,719][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:21:39,218][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:21:39,718][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:21:40,219][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:21:40,719][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:21:41,218][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:21:41,718][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:21:42,217][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:21:42,717][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:21:43,217][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:21:43,715][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:21:44,215][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:21:44,713][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:21:45,211][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:21:45,710][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:21:46,209][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:21:46,709][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:21:47,208][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:21:47,707][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:21:48,207][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:21:48,707][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:21:49,206][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:21:49,706][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:21:50,206][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:21:50,706][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:21:51,205][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:21:51,704][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:21:52,204][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:21:52,703][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:21:53,204][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:21:53,703][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:21:54,203][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:21:54,701][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:21:55,201][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:21:55,701][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:21:56,199][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:21:56,700][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:21:57,199][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:21:57,699][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:21:58,199][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:21:58,700][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:21:59,201][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:21:59,701][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:22:00,201][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:22:00,700][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:22:01,201][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:22:01,699][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:22:02,200][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:22:02,701][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:22:03,200][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:22:03,699][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:22:04,199][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:22:04,698][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:22:05,197][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:22:05,696][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:22:06,195][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:22:06,694][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:22:07,194][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:22:07,693][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:22:08,192][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:22:08,692][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:22:09,191][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:22:09,690][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:22:10,188][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:22:10,686][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:22:11,185][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:22:11,685][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:22:12,184][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:22:12,683][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:22:13,182][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:22:13,681][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:22:14,180][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:22:14,679][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:22:15,179][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:22:15,678][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:22:16,178][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:22:16,677][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:22:17,176][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:22:17,675][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:22:18,173][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:22:18,672][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:22:19,172][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:22:19,671][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:22:20,170][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:22:20,669][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:22:21,168][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:22:21,667][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:22:22,166][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:22:22,665][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:22:23,164][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:22:23,664][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:22:24,164][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:22:24,663][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:22:25,164][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:22:25,663][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:22:26,162][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:22:26,660][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:22:27,159][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:22:27,658][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:22:28,157][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:22:28,656][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:22:29,155][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:22:29,654][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:22:30,152][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:22:30,652][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:22:31,152][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:22:31,651][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:22:32,151][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:22:32,650][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:22:33,149][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:22:33,648][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:22:34,147][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:22:34,646][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:22:35,146][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:22:35,645][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:22:36,145][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:22:36,644][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:22:37,145][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 06:22:37,779][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 06:22:38,596][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:22:38,599][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:22:38,600][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:22:39,257][__main__][INFO] - Iteration 464 took 1m 14s (9.49% Gen, 89.63% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 22m 15s. Estimated total time: 62h 10m 12s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 20s, 500 more iterations: 10h 21m 42s. [2026-03-26 06:22:39,260][__main__][INFO] - Starting iteration 464. [2026-03-26 06:22:39,670][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:22:39,670][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:22:43,538][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:22:46,326][__main__][INFO] - Number of regex retries in iteration 464: 1 [2026-03-26 06:22:46,327][__main__][INFO] - agents played in iteration 464 are Alice, Bob [2026-03-26 06:22:47,223][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:22:47,268][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:22:47,269][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:22:47,269][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:22:47,880][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:22:48,342][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:22:48,847][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:22:49,347][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:22:49,847][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:22:50,347][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:22:50,846][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:22:51,345][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:22:51,845][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:22:52,347][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:22:52,847][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:22:53,347][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:22:53,847][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:22:54,347][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:22:54,848][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:22:55,348][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:22:55,848][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:22:56,347][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:22:56,847][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:22:57,346][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:22:57,846][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:22:58,346][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:22:58,845][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:22:59,344][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:22:59,844][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:23:00,344][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:23:00,844][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:23:01,343][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:23:01,842][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:23:02,341][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:23:02,840][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:23:03,340][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:23:03,839][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:23:04,338][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:23:04,836][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:23:05,334][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:23:05,832][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:23:06,330][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:23:06,828][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:23:07,326][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:23:07,824][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:23:08,323][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:23:08,822][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:23:09,322][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:23:09,821][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:23:10,321][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:23:10,821][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:23:11,320][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:23:11,820][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:23:12,318][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:23:12,817][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:23:13,316][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:23:13,813][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:23:14,311][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:23:14,808][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:23:15,306][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:23:15,805][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:23:16,304][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:23:16,803][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:23:17,302][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:23:17,801][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:23:18,302][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:23:18,800][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:23:19,299][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:23:19,798][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:23:20,298][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:23:20,798][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:23:21,297][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:23:21,796][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:23:22,295][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:23:22,795][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:23:23,295][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:23:23,795][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:23:24,296][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:23:24,795][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:23:25,296][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:23:25,800][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:23:26,301][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:23:26,802][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:23:27,303][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:23:27,803][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:23:28,304][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:23:28,804][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:23:29,304][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:23:29,808][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:23:30,310][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:23:30,810][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:23:31,311][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:23:31,811][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:23:32,311][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:23:32,811][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:23:33,311][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:23:33,811][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:23:34,313][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:23:34,814][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:23:35,315][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:23:35,816][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:23:36,316][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:23:36,816][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:23:37,317][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:23:37,816][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:23:38,318][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:23:38,821][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:23:39,322][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:23:39,823][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:23:40,325][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:23:40,826][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:23:41,328][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:23:41,829][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:23:42,333][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:23:42,835][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:23:43,337][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:23:43,840][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:23:44,341][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:23:44,842][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:23:45,344][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:23:45,847][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:23:46,354][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:23:46,856][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:23:47,357][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:23:47,862][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:23:48,366][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:23:48,867][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:23:49,368][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:23:49,869][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:23:50,376][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:23:50,879][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:23:51,381][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:23:51,884][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:23:52,541][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 06:23:53,288][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:23:53,291][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:23:53,292][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:23:53,974][__main__][INFO] - Iteration 465 took 1m 14s (8.96% Gen, 90.12% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 6m 4s. Estimated total time: 61h 55m 16s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 50s, 500 more iterations: 10h 19m 12s. [2026-03-26 06:23:53,976][__main__][INFO] - Starting iteration 465. [2026-03-26 06:23:54,375][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:23:54,376][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:24:01,403][__main__][INFO] - Number of regex retries in iteration 465: 0 [2026-03-26 06:24:01,404][__main__][INFO] - agents played in iteration 465 are Alice, Bob [2026-03-26 06:24:02,288][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:24:02,334][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:24:02,334][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:24:02,335][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:24:02,949][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:24:03,407][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:24:03,911][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:24:04,411][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:24:04,910][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:24:05,409][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:24:05,909][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:24:06,408][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:24:06,908][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:24:07,407][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:24:07,906][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:24:08,405][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:24:08,904][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:24:09,402][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:24:09,900][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:24:10,400][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:24:10,899][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:24:11,399][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:24:11,898][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:24:12,396][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:24:12,894][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:24:13,392][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:24:13,890][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:24:14,387][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:24:14,885][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:24:15,385][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:24:15,884][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:24:16,383][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:24:16,883][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:24:17,383][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:24:17,883][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:24:18,386][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:24:18,886][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:24:19,386][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:24:19,885][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:24:20,384][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:24:20,883][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:24:21,382][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:24:21,881][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:24:22,380][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:24:22,879][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:24:23,381][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:24:23,884][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:24:24,385][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:24:24,887][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:24:25,389][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:24:25,889][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:24:26,388][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:24:26,888][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:24:27,387][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:24:27,885][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:24:28,385][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:24:28,885][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:24:29,385][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:24:29,884][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:24:30,383][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:24:30,883][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:24:31,382][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:24:31,883][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:24:32,382][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:24:32,882][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:24:33,381][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:24:33,880][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:24:34,379][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:24:34,878][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:24:35,378][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:24:35,880][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:24:36,379][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:24:36,879][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:24:37,378][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:24:37,877][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:24:38,376][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:24:38,874][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:24:39,372][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:24:39,869][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:24:40,367][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:24:40,864][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:24:41,362][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:24:41,860][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:24:42,357][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:24:42,855][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:24:43,352][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:24:43,849][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:24:44,347][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:24:44,847][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:24:45,345][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:24:45,844][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:24:46,343][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:24:46,841][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:24:47,340][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:24:47,840][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:24:48,341][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:24:48,846][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:24:49,347][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:24:49,847][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:24:50,348][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:24:50,848][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:24:51,348][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:24:51,849][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:24:52,348][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:24:52,850][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:24:53,350][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:24:53,850][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:24:54,350][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:24:54,851][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:24:55,353][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:24:55,855][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:24:56,357][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:24:56,859][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:24:57,361][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:24:57,864][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:24:58,367][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:24:58,869][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:24:59,370][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:24:59,870][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:25:00,370][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:25:00,875][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:25:01,377][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:25:01,878][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:25:02,379][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:25:02,880][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:25:03,383][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:25:03,885][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:25:04,387][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:25:04,892][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:25:05,394][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:25:05,896][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:25:06,399][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:25:06,902][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 06:25:07,539][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 06:25:14,615][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:25:14,617][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:25:14,619][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:25:16,093][__main__][INFO] - Iteration 466 took 1m 21s (8.60% Gen, 89.59% Train). Generation: 7s, Training: 1m 13s. Estimated remaining time: 58h 15m 21s. Estimated total time: 68h 5m 55s. Time estimates for 10 more iterations: 13m 37s, 100 more iterations: 2h 16m 11s, 500 more iterations: 11h 20m 59s. [2026-03-26 06:25:16,095][__main__][INFO] - Starting iteration 466. [2026-03-26 06:25:16,495][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:25:16,496][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:25:17,095][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:25:18,582][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:25:19,291][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:25:23,443][__main__][INFO] - Number of regex retries in iteration 466: 3 [2026-03-26 06:25:23,444][__main__][INFO] - agents played in iteration 466 are Alice, Bob [2026-03-26 06:25:24,338][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:25:24,384][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:25:24,384][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:25:24,385][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:25:24,994][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:25:25,449][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:25:25,955][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:25:26,455][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:25:26,953][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:25:27,452][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:25:27,950][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:25:28,448][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:25:28,947][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:25:29,444][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:25:29,942][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:25:30,441][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:25:30,938][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:25:31,436][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:25:31,933][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:25:32,430][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:25:32,928][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:25:33,430][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:25:33,931][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:25:34,430][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:25:34,929][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:25:35,429][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:25:35,928][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:25:36,427][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:25:36,927][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:25:37,427][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:25:37,925][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:25:38,425][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:25:38,925][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:25:39,425][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:25:39,925][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:25:40,424][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:25:40,924][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:25:41,422][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:25:41,920][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:25:42,418][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:25:42,915][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:25:43,413][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:25:43,910][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:25:44,407][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:25:44,906][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:25:45,408][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:25:45,908][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:25:46,408][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:25:46,908][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:25:47,408][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:25:47,909][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:25:48,411][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:25:48,913][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:25:49,413][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:25:49,913][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:25:50,412][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:25:50,912][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:25:51,412][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:25:51,912][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:25:52,412][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:25:52,912][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:25:53,411][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:25:53,910][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:25:54,409][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:25:54,909][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:25:55,410][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:25:55,910][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:25:56,409][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:25:56,908][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:25:57,407][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:25:57,905][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:25:58,403][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:25:58,902][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:25:59,401][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:25:59,900][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:26:00,399][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:26:00,897][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:26:01,393][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:26:01,890][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:26:02,387][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:26:02,885][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:26:03,383][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:26:03,881][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:26:04,378][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:26:04,875][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:26:05,374][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:26:05,874][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:26:06,373][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:26:06,873][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:26:07,372][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:26:07,871][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:26:08,371][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:26:08,871][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:26:09,369][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:26:09,868][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:26:10,366][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:26:10,865][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:26:11,364][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:26:11,864][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:26:12,363][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:26:12,862][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:26:13,360][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:26:13,859][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:26:14,357][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:26:14,857][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:26:15,357][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:26:15,855][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:26:16,354][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:26:16,853][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:26:17,352][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:26:17,852][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:26:18,354][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:26:18,855][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:26:19,355][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:26:19,855][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:26:20,355][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:26:20,855][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:26:21,355][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:26:21,856][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:26:22,357][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:26:22,858][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:26:23,358][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:26:23,858][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:26:24,358][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:26:24,858][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:26:25,360][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:26:25,860][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:26:26,360][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:26:26,860][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:26:27,359][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:26:27,857][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:26:28,672][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:26:29,243][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 06:26:29,956][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.29%, ΔTime: 00:01:04 [2026-03-26 06:26:30,809][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:26:30,811][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:26:30,813][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:26:31,531][__main__][INFO] - Iteration 467 took 1m 15s (9.26% Gen, 89.78% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 52h 40m 0s. Estimated total time: 62h 31m 50s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 3s, 500 more iterations: 10h 25m 18s. [2026-03-26 06:26:31,533][__main__][INFO] - Starting iteration 467. [2026-03-26 06:26:31,935][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:26:31,936][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:26:35,048][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:26:39,189][__main__][INFO] - Number of regex retries in iteration 467: 1 [2026-03-26 06:26:39,189][__main__][INFO] - agents played in iteration 467 are Alice, Bob [2026-03-26 06:26:40,064][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:26:40,109][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:26:40,110][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:26:40,110][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:26:40,714][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:26:41,168][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:26:41,686][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:26:42,171][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:26:42,671][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:26:43,169][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:26:43,668][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:26:44,166][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:26:44,666][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:26:45,167][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:26:45,670][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:26:46,171][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:26:46,670][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:26:47,168][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:26:47,667][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:26:48,167][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:26:48,667][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:26:49,167][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:26:49,668][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:26:50,167][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:26:50,667][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:26:51,166][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:26:51,666][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:26:52,165][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:26:52,665][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:26:53,165][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:26:53,665][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:26:54,165][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:26:54,664][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:26:55,165][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:26:55,664][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:26:56,164][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:26:56,664][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:26:57,183][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:26:57,687][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:26:58,188][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:26:58,689][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:26:59,190][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:26:59,690][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:27:00,190][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:27:00,691][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:27:01,190][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:27:01,690][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:27:02,190][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:27:02,689][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:27:03,190][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:27:03,690][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:27:04,189][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:27:04,688][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:27:05,188][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:27:05,688][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:27:06,186][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:27:06,685][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:27:07,184][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:27:07,684][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:27:08,183][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:27:08,682][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:27:09,184][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:27:09,685][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:27:10,185][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:27:10,685][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:27:11,185][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:27:11,685][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:27:12,186][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:27:12,686][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:27:13,186][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:27:13,684][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:27:14,182][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:27:14,680][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:27:15,178][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:27:15,676][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:27:16,174][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:27:16,672][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:27:17,171][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:27:17,670][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:27:18,170][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:27:18,669][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:27:19,168][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:27:19,667][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:27:20,166][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:27:20,664][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:27:21,164][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:27:21,663][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:27:22,162][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:27:22,661][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:27:23,163][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:27:23,662][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:27:24,162][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:27:24,662][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:27:25,162][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:27:25,661][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:27:26,160][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:27:26,659][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:27:27,158][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:27:27,659][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:27:28,160][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:27:28,660][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:27:29,160][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:27:29,661][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:27:30,163][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:27:30,663][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:27:31,165][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:27:31,666][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:27:32,167][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:27:32,668][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:27:33,168][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:27:33,667][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:27:34,168][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:27:34,669][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:27:35,169][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:27:35,669][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:27:36,169][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:27:36,669][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:27:37,168][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:27:37,666][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:27:38,166][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:27:38,665][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:27:39,164][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:27:39,662][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:27:40,161][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:27:40,660][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:27:41,159][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:27:41,661][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:27:42,162][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:27:42,662][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:27:43,162][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:27:43,662][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:27:44,162][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:27:44,663][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 06:27:45,364][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 06:27:46,008][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:27:46,012][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:27:46,014][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:27:46,618][__main__][INFO] - Iteration 468 took 1m 14s (9.71% Gen, 89.48% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 21m 5s. Estimated total time: 62h 14m 10s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 28s, 500 more iterations: 10h 22m 21s. [2026-03-26 06:27:46,621][__main__][INFO] - Starting iteration 468. [2026-03-26 06:27:47,018][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:27:47,019][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:27:53,442][__main__][INFO] - Number of regex retries in iteration 468: 0 [2026-03-26 06:27:53,443][__main__][INFO] - agents played in iteration 468 are Alice, Bob [2026-03-26 06:27:54,383][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:27:54,428][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:27:54,428][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:27:54,429][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:27:55,039][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:27:55,494][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:27:55,999][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:27:56,498][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:27:56,997][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:27:57,496][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:27:57,995][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:27:58,493][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:27:58,992][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:27:59,490][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:27:59,987][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:28:00,485][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:28:00,984][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:28:01,482][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:28:01,981][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:28:02,478][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:28:02,976][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:28:03,476][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:28:03,974][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:28:04,473][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:28:04,972][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:28:05,471][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:28:05,970][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:28:06,468][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:28:06,967][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:28:07,467][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:28:07,966][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:28:08,464][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:28:08,962][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:28:09,461][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:28:09,960][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:28:10,458][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:28:10,956][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:28:11,455][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:28:11,953][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:28:12,452][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:28:12,951][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:28:13,451][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:28:13,950][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:28:14,448][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:28:14,947][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:28:15,452][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:28:15,955][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:28:16,455][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:28:16,956][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:28:17,456][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:28:17,957][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:28:18,459][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:28:18,960][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:28:19,461][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:28:19,960][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:28:20,459][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:28:20,958][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:28:21,456][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:28:21,954][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:28:22,453][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:28:22,951][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:28:23,452][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:28:23,952][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:28:24,454][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:28:24,954][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:28:25,455][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:28:25,954][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:28:26,455][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:28:26,954][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:28:27,454][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:28:27,953][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:28:28,451][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:28:28,950][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:28:29,449][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:28:29,948][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:28:30,447][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:28:30,946][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:28:31,444][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:28:31,943][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:28:32,441][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:28:32,939][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:28:33,437][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:28:33,934][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:28:34,431][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:28:34,930][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:28:35,429][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:28:35,929][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:28:36,427][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:28:36,925][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:28:37,424][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:28:37,924][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:28:38,423][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:28:38,922][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:28:39,421][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:28:39,920][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:28:40,420][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:28:40,918][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:28:41,416][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:28:41,914][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:28:42,413][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:28:42,911][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:28:43,410][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:28:43,909][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:28:44,407][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:28:44,906][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:28:45,404][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:28:45,903][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:28:46,402][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:28:46,902][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:28:47,403][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:28:47,905][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:28:48,407][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:28:48,908][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:28:49,408][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:28:49,910][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:28:50,409][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:28:50,910][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:28:51,409][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:28:51,908][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:28:52,406][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:28:52,906][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:28:53,406][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:28:53,906][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:28:54,405][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:28:54,904][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:28:55,405][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:28:55,905][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:28:56,405][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:28:56,906][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:28:57,408][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:28:57,909][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:28:58,409][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:28:58,908][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 06:28:59,553][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:29:00,286][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:29:00,288][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:29:00,290][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:29:00,939][__main__][INFO] - Iteration 469 took 1m 13s (8.69% Gen, 90.43% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 41m 45s. Estimated total time: 61h 36m 4s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 12s, 500 more iterations: 10h 16m 0s. [2026-03-26 06:29:00,941][__main__][INFO] - Starting iteration 469. [2026-03-26 06:29:01,344][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:29:01,344][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:29:01,925][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:29:08,150][__main__][INFO] - Number of regex retries in iteration 469: 1 [2026-03-26 06:29:08,151][__main__][INFO] - agents played in iteration 469 are Alice, Bob [2026-03-26 06:29:09,036][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:29:09,081][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:29:09,082][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:29:09,082][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:29:09,694][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:29:10,154][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:29:10,659][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:29:11,158][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:29:11,659][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:29:12,158][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:29:12,659][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:29:13,160][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:29:13,661][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:29:14,161][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:29:14,661][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:29:15,160][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:29:15,659][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:29:16,161][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:29:16,661][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:29:17,160][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:29:17,659][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:29:18,160][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:29:18,659][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:29:19,157][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:29:19,656][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:29:20,154][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:29:20,653][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:29:21,151][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:29:21,651][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:29:22,149][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:29:22,646][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:29:23,144][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:29:23,644][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:29:24,142][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:29:24,641][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:29:25,140][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:29:25,640][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:29:26,140][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:29:26,639][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:29:27,137][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:29:27,637][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:29:28,137][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:29:28,636][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:29:29,135][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:29:29,635][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:29:30,136][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:29:30,635][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:29:31,134][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:29:31,634][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:29:32,134][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:29:32,633][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:29:33,133][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:29:33,633][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:29:34,132][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:29:34,631][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:29:35,130][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:29:35,629][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:29:36,129][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:29:36,628][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:29:37,128][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:29:37,627][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:29:38,127][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:29:38,627][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:29:39,127][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:29:39,627][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:29:40,127][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:29:40,626][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:29:41,126][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:29:41,626][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:29:42,127][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:29:42,628][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:29:43,129][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:29:43,630][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:29:44,130][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:29:44,631][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:29:45,131][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:29:45,635][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:29:46,137][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:29:46,638][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:29:47,137][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:29:47,637][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:29:48,136][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:29:48,638][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:29:49,138][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:29:49,638][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:29:50,137][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:29:50,636][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:29:51,135][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:29:51,634][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:29:52,132][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:29:52,631][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:29:53,131][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:29:53,630][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:29:54,128][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:29:54,627][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:29:55,127][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:29:55,626][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:29:56,124][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:29:56,624][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:29:57,122][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:29:57,621][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:29:58,121][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:29:58,619][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:29:59,118][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:29:59,616][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:30:00,116][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:30:00,614][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:30:01,113][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:30:01,612][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:30:02,112][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:30:02,611][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:30:03,111][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:30:03,610][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:30:04,109][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:30:04,610][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:30:05,111][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:30:05,610][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:30:06,109][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:30:06,608][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:30:07,107][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:30:07,607][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:30:08,107][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:30:08,606][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:30:09,105][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:30:09,604][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:30:10,105][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:30:10,604][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:30:11,104][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:30:11,603][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:30:12,103][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:30:12,603][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:30:13,101][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:30:13,601][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 06:30:14,235][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:30:14,923][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:30:14,925][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:30:14,927][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:30:15,835][__main__][INFO] - Iteration 470 took 1m 14s (9.14% Gen, 89.64% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 52h 9m 1s. Estimated total time: 62h 4m 35s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 45s. [2026-03-26 06:30:15,837][__main__][INFO] - Starting iteration 470. [2026-03-26 06:30:16,237][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:30:16,238][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:30:16,815][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:30:23,523][__main__][INFO] - Number of regex retries in iteration 470: 1 [2026-03-26 06:30:23,524][__main__][INFO] - agents played in iteration 470 are Alice, Bob [2026-03-26 06:30:24,406][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:30:24,451][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:30:24,451][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:30:24,452][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:30:25,072][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:30:25,530][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:30:26,036][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:30:26,538][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:30:27,040][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:30:27,541][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:30:28,041][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:30:28,543][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:30:29,044][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:30:29,544][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:30:30,045][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:30:30,543][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:30:31,043][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:30:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:30:32,043][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:30:32,543][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:30:33,041][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:30:33,539][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:30:34,038][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:30:34,535][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:30:35,033][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:30:35,531][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:30:36,030][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:30:36,528][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:30:37,026][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:30:37,524][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:30:38,023][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:30:38,522][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:30:39,022][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:30:39,521][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:30:40,022][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:30:40,521][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:30:41,020][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:30:41,520][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:30:42,019][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:30:42,518][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:30:43,017][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:30:43,516][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:30:44,016][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:30:44,516][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:30:45,015][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:30:45,515][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:30:46,015][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:30:46,513][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:30:47,012][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:30:47,511][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:30:48,012][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:30:48,514][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:30:49,014][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:30:49,513][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:30:50,013][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:30:50,512][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:30:51,011][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:30:51,511][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:30:52,011][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:30:52,510][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:30:53,009][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:30:53,509][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:30:54,008][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:30:54,509][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:30:55,009][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:30:55,510][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:30:56,010][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:30:56,510][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:30:57,009][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:30:57,509][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:30:58,010][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:30:58,511][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:30:59,012][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:30:59,511][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:31:00,011][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:31:00,511][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:31:01,010][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:31:01,510][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:31:02,011][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:31:02,511][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:31:03,011][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:31:03,510][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:31:04,010][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:31:04,509][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:31:05,009][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:31:05,507][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:31:06,005][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:31:06,504][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:31:07,003][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:31:07,501][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:31:07,999][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:31:08,497][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:31:08,994][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:31:09,493][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:31:09,991][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:31:10,490][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:31:10,989][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:31:11,489][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:31:11,987][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:31:12,486][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:31:12,985][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:31:13,485][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:31:13,985][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:31:14,485][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:31:14,984][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:31:15,487][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:31:15,989][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:31:16,489][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:31:16,990][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:31:17,491][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:31:17,992][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:31:18,494][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:31:18,993][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:31:19,491][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:31:19,989][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:31:20,489][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:31:20,988][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:31:21,488][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:31:21,988][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:31:22,488][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:31:22,987][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:31:23,488][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:31:23,989][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:31:24,489][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:31:24,989][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:31:25,490][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:31:25,990][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:31:26,490][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:31:26,989][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:31:27,489][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:31:27,991][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:31:28,490][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:31:28,990][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 06:31:29,647][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 06:31:30,434][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:31:30,436][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:31:30,438][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:31:31,223][__main__][INFO] - Iteration 471 took 1m 14s (9.72% Gen, 89.24% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 32m 28s. Estimated total time: 62h 29m 17s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 58s, 500 more iterations: 10h 24m 52s. [2026-03-26 06:31:31,225][__main__][INFO] - Starting iteration 471. [2026-03-26 06:31:31,629][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:31:31,629][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:31:33,986][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:31:38,710][__main__][INFO] - Number of regex retries in iteration 471: 1 [2026-03-26 06:31:38,711][__main__][INFO] - agents played in iteration 471 are Alice, Bob [2026-03-26 06:31:39,599][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:31:39,644][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:31:39,645][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:31:39,646][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:31:40,269][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:31:40,725][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:31:41,229][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:31:41,730][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:31:42,230][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:31:42,729][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:31:43,229][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:31:43,730][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:31:44,230][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:31:44,732][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:31:45,233][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:31:45,733][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:31:46,233][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:31:46,734][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:31:47,234][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:31:47,735][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:31:48,239][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:31:48,739][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:31:49,239][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:31:49,739][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:31:50,240][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:31:50,740][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:31:51,240][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:31:51,741][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:31:52,241][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:31:52,740][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:31:53,240][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:31:53,740][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:31:54,240][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:31:54,740][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:31:55,240][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:31:55,739][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:31:56,238][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:31:56,741][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:31:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:31:57,744][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:31:58,246][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:31:58,748][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:31:59,250][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:31:59,752][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:32:00,253][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:32:00,753][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:32:01,253][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:32:01,754][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:32:02,255][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:32:02,755][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:32:03,255][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:32:03,755][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:32:04,255][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:32:04,754][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:32:05,253][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:32:05,753][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:32:06,252][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:32:06,751][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:32:07,250][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:32:07,751][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:32:08,250][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:32:08,749][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:32:09,247][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:32:09,746][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:32:10,244][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:32:10,743][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:32:11,242][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:32:11,742][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:32:12,241][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:32:12,741][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:32:13,240][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:32:13,738][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:32:14,236][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:32:14,735][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:32:15,233][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:32:15,733][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:32:16,232][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:32:16,732][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:32:17,231][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:32:17,731][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:32:18,234][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:32:18,735][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:32:19,236][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:32:19,736][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:32:20,236][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:32:20,735][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:32:21,235][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:32:21,734][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:32:22,234][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:32:22,733][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:32:23,234][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:32:23,733][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:32:24,233][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:32:24,733][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:32:25,233][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:32:25,732][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:32:26,232][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:32:26,731][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:32:27,230][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:32:27,729][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:32:28,229][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:32:28,731][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:32:29,234][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:32:29,735][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:32:30,236][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:32:30,777][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:32:31,277][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:32:31,780][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:32:32,281][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:32:32,781][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:32:33,282][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:32:33,783][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:32:34,283][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:32:34,783][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:32:35,283][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:32:35,783][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:32:36,283][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:32:36,784][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:32:37,283][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:32:37,781][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:32:38,280][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:32:38,779][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:32:39,278][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:32:39,775][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:32:40,273][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:32:40,771][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:32:41,270][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:32:41,769][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:32:42,268][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:32:42,768][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:32:43,267][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:32:43,766][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:32:44,266][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21750 tokens. [2026-03-26 06:32:44,919][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 06:32:45,655][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:32:45,658][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:32:45,659][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:32:46,410][__main__][INFO] - Iteration 472 took 1m 14s (9.47% Gen, 89.52% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 21m 0s. Estimated total time: 62h 19m 4s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 38s, 500 more iterations: 10h 23m 10s. [2026-03-26 06:32:46,413][__main__][INFO] - Starting iteration 472. [2026-03-26 06:32:46,816][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:32:46,817][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:32:53,746][__main__][INFO] - Number of regex retries in iteration 472: 0 [2026-03-26 06:32:53,747][__main__][INFO] - agents played in iteration 472 are Alice, Bob [2026-03-26 06:32:54,621][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:32:54,666][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:32:54,667][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:32:54,667][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:32:55,272][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:32:55,728][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:32:56,233][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:32:56,734][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:32:57,233][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:32:57,734][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:32:58,234][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:32:58,734][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:32:59,234][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:32:59,732][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:33:00,231][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:33:00,731][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:33:01,230][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:33:01,728][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:33:02,227][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:33:02,726][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:33:03,225][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:33:03,725][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:33:04,225][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:33:04,724][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:33:05,223][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:33:05,722][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:33:06,222][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:33:06,722][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:33:07,221][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:33:07,721][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:33:08,221][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:33:08,721][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:33:09,220][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:33:09,719][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:33:10,218][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:33:10,715][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:33:11,214][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:33:11,712][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:33:12,211][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:33:12,709][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:33:13,208][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:33:13,706][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:33:14,205][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:33:14,704][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:33:15,203][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:33:15,704][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:33:16,204][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:33:16,703][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:33:17,203][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:33:17,701][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:33:18,202][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:33:18,702][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:33:19,201][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:33:19,703][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:33:20,203][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:33:20,703][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:33:21,202][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:33:21,702][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:33:22,202][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:33:22,703][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:33:23,203][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:33:23,702][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:33:24,202][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:33:24,699][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:33:25,198][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:33:25,696][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:33:26,194][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:33:26,692][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:33:27,191][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:33:27,690][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:33:28,189][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:33:28,690][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:33:29,191][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:33:29,690][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:33:30,190][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:33:30,692][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:33:31,191][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:33:31,689][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:33:32,188][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:33:32,686][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:33:33,184][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:33:33,682][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:33:34,183][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:33:34,683][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:33:35,181][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:33:35,680][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:33:36,179][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:33:36,677][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:33:37,176][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:33:37,674][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:33:38,173][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:33:38,671][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:33:39,170][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:33:39,669][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:33:40,167][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:33:40,665][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:33:41,164][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:33:41,663][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:33:42,164][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:33:42,663][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:33:43,163][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:33:43,661][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:33:44,160][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:33:44,657][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:33:45,155][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:33:45,654][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:33:46,152][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:33:46,650][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:33:47,149][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:33:47,647][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:33:48,148][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:33:48,649][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:33:49,148][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:33:49,648][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:33:50,148][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:33:50,647][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:33:51,146][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:33:51,646][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:33:52,146][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:33:52,646][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:33:53,146][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:33:53,646][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:33:54,146][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:33:54,646][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:33:55,147][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:33:55,645][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:33:56,144][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:33:56,643][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:33:57,142][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:33:57,642][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:33:58,141][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:33:58,640][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:33:59,138][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 06:33:59,766][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 06:34:00,507][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:34:00,509][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:34:00,511][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:34:01,163][__main__][INFO] - Iteration 473 took 1m 14s (9.32% Gen, 89.80% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 58m 10s. Estimated total time: 61h 57m 29s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 54s, 500 more iterations: 10h 19m 34s. [2026-03-26 06:34:01,165][__main__][INFO] - Starting iteration 473. [2026-03-26 06:34:01,563][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:34:01,563][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:34:08,920][__main__][INFO] - Number of regex retries in iteration 473: 0 [2026-03-26 06:34:08,920][__main__][INFO] - agents played in iteration 473 are Alice, Bob [2026-03-26 06:34:09,801][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:34:09,846][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:34:09,847][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:34:09,848][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:34:10,448][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:34:10,938][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:34:11,442][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:34:11,942][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:34:12,441][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:34:12,940][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:34:13,438][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:34:13,938][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:34:14,438][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:34:14,937][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:34:15,439][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:34:15,940][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:34:16,439][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:34:16,937][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:34:17,435][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:34:17,934][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:34:18,435][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:34:18,934][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:34:19,433][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:34:19,931][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:34:20,430][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:34:20,928][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:34:21,428][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:34:21,926][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:34:22,425][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:34:22,926][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:34:23,427][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:34:23,928][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:34:24,428][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:34:24,929][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:34:25,431][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:34:25,932][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:34:26,432][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:34:26,932][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:34:27,432][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:34:27,932][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:34:28,433][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:34:28,933][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:34:29,434][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:34:29,933][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:34:30,434][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:34:30,931][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:34:31,430][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:34:31,930][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:34:32,430][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:34:32,928][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:34:33,426][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:34:33,925][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:34:34,423][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:34:34,922][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:34:35,423][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:34:35,923][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:34:36,424][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:34:36,924][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:34:37,424][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:34:37,923][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:34:38,424][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:34:38,923][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:34:39,423][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:34:39,923][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:34:40,423][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:34:40,923][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:34:41,423][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:34:41,922][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:34:42,422][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:34:42,920][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:34:43,419][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:34:43,918][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:34:44,417][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:34:44,916][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:34:45,414][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:34:45,912][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:34:46,411][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:34:46,909][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:34:47,408][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:34:47,907][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:34:48,407][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:34:48,905][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:34:49,402][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:34:49,900][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:34:50,397][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:34:50,894][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:34:51,392][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:34:51,891][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:34:52,390][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:34:52,888][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:34:53,387][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:34:53,885][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:34:54,382][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:34:54,882][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:34:55,385][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:34:55,885][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:34:56,385][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:34:56,887][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:34:57,387][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:34:57,888][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:34:58,388][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:34:58,889][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:34:59,390][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:34:59,889][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:35:00,390][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:35:00,889][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:35:01,389][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:35:01,889][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:35:02,389][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:35:02,888][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:35:03,385][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:35:03,882][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:35:04,380][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:35:04,877][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:35:05,375][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:35:05,873][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:35:06,371][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:35:06,871][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:35:07,371][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:35:07,870][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:35:08,369][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:35:08,869][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:35:09,368][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:35:09,868][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:35:10,367][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:35:10,867][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:35:11,366][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:35:11,864][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:35:12,363][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:35:12,862][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:35:13,360][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:35:13,860][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:35:14,359][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-26 06:35:14,985][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 06:35:15,722][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:35:15,725][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:35:15,726][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:35:16,386][__main__][INFO] - Iteration 474 took 1m 14s (9.83% Gen, 89.28% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 20m 36s. Estimated total time: 62h 21m 10s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 42s, 500 more iterations: 10h 23m 31s. [2026-03-26 06:35:16,388][__main__][INFO] - Starting iteration 474. [2026-03-26 06:35:16,787][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:35:16,788][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:35:18,595][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:35:23,831][__main__][INFO] - Number of regex retries in iteration 474: 1 [2026-03-26 06:35:23,832][__main__][INFO] - agents played in iteration 474 are Alice, Bob [2026-03-26 06:35:24,706][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:35:24,751][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:35:24,752][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:35:24,753][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:35:25,357][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:35:25,810][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:35:26,313][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:35:26,813][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:35:27,312][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:35:27,811][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:35:28,310][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:35:28,809][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:35:29,308][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:35:29,807][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:35:30,304][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:35:30,805][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:35:31,303][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:35:31,802][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:35:32,300][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:35:32,800][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:35:33,299][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:35:33,797][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:35:34,294][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:35:34,792][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:35:35,290][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:35:35,788][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:35:36,285][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:35:36,783][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:35:37,280][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:35:37,778][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:35:38,276][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:35:38,775][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:35:39,274][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:35:39,772][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:35:40,271][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:35:40,769][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:35:41,268][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:35:41,766][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:35:42,264][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:35:42,762][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:35:43,261][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:35:43,760][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:35:44,259][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:35:44,757][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:35:45,256][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:35:45,757][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:35:46,256][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:35:46,755][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:35:47,253][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:35:47,752][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:35:48,253][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:35:48,752][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:35:49,251][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:35:49,749][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:35:50,248][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:35:50,746][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:35:51,246][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:35:51,745][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:35:52,244][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:35:52,742][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:35:53,240][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:35:53,741][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:35:54,240][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:35:54,741][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:35:55,242][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:35:55,743][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:35:56,244][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:35:56,744][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:35:57,244][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:35:57,743][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:35:58,243][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:35:58,743][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:35:59,243][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:35:59,743][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:36:00,242][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:36:00,741][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:36:01,239][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:36:01,738][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:36:02,236][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:36:02,734][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:36:03,233][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:36:03,731][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:36:04,229][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:36:04,728][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:36:05,227][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:36:05,726][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:36:06,226][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:36:06,726][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:36:07,224][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:36:07,723][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:36:08,221][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:36:08,720][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:36:09,217][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:36:09,715][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:36:10,213][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:36:10,712][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:36:11,211][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:36:11,709][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:36:12,208][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:36:12,707][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:36:13,206][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:36:13,705][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:36:14,204][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:36:14,703][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:36:15,201][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:36:15,699][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:36:16,196][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:36:16,694][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:36:17,191][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:36:17,688][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:36:18,189][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:36:18,688][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:36:19,186][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:36:19,684][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:36:20,182][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:36:20,680][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:36:21,177][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:36:21,676][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:36:22,174][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:36:22,672][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:36:23,171][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:36:23,672][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:36:24,171][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:36:24,670][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:36:25,169][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:36:25,669][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:36:26,170][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:36:26,669][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:36:27,169][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:36:27,669][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:36:28,169][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:36:28,669][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:36:29,168][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:36:29,799][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 06:36:30,643][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:36:30,645][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:36:30,647][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:36:31,335][__main__][INFO] - Iteration 475 took 1m 14s (9.45% Gen, 89.63% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 5m 35s. Estimated total time: 62h 7m 24s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 14s, 500 more iterations: 10h 21m 14s. [2026-03-26 06:36:31,337][__main__][INFO] - Starting iteration 475. [2026-03-26 06:36:31,738][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:36:31,739][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:36:38,471][__main__][INFO] - Number of regex retries in iteration 475: 0 [2026-03-26 06:36:38,472][__main__][INFO] - agents played in iteration 475 are Alice, Bob [2026-03-26 06:36:39,349][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:36:39,394][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:36:39,395][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:36:39,395][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:36:40,004][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:36:40,456][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:36:40,957][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:36:41,453][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:36:41,948][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:36:42,444][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:36:42,938][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:36:43,433][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:36:43,929][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:36:44,429][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:36:44,927][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:36:45,427][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:36:45,926][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:36:46,424][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:36:46,923][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:36:47,421][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:36:47,920][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:36:48,420][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:36:48,919][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:36:49,419][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:36:49,918][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:36:50,417][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:36:50,915][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:36:51,414][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:36:51,913][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:36:52,411][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:36:52,910][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:36:53,410][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:36:53,910][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:36:54,409][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:36:54,907][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:36:55,406][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:36:55,907][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:36:56,407][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:36:56,907][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:36:57,407][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:36:57,907][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:36:58,407][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:36:58,906][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:36:59,406][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:36:59,906][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:37:00,407][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:37:00,907][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:37:01,407][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:37:01,906][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:37:02,406][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:37:02,906][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:37:03,406][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:37:03,907][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:37:04,406][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:37:04,906][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:37:05,404][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:37:05,904][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:37:06,403][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:37:06,903][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:37:07,404][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:37:07,905][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:37:08,404][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:37:08,903][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:37:09,401][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:37:09,900][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:37:10,398][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:37:10,897][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:37:11,396][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:37:11,894][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:37:12,389][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:37:12,884][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:37:13,380][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:37:13,874][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:37:14,369][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:37:14,863][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:37:15,360][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:37:15,858][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:37:16,356][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:37:16,855][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:37:17,353][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:37:17,853][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:37:18,354][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:37:18,853][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:37:19,352][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:37:19,851][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:37:20,351][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:37:20,850][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:37:21,349][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:37:21,849][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:37:22,348][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:37:22,846][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:37:23,345][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:37:23,845][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:37:24,345][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:37:24,845][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:37:25,345][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:37:25,844][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:37:26,342][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:37:26,841][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:37:27,339][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:37:27,838][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:37:28,339][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:37:28,839][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:37:29,339][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:37:29,839][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:37:30,338][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:37:30,837][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:37:31,337][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:37:31,837][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:37:32,336][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:37:32,836][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:37:33,336][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:37:33,835][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:37:34,335][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:37:34,834][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:37:35,334][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:37:35,835][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:37:36,335][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:37:36,835][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:37:37,335][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:37:37,834][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:37:38,332][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:37:38,833][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:37:39,333][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:37:39,833][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:37:40,332][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:37:40,834][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:37:41,333][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:37:41,832][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:37:42,332][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:37:42,833][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:37:43,333][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:37:43,833][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:37:44,478][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.27%, ΔTime: 00:01:04 [2026-03-26 06:37:45,228][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:37:45,230][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:37:45,232][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:37:45,893][__main__][INFO] - Iteration 476 took 1m 14s (9.08% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 44m 40s. Estimated total time: 61h 47m 44s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 57s. [2026-03-26 06:37:45,895][__main__][INFO] - Starting iteration 476. [2026-03-26 06:37:46,299][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:37:46,299][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:37:50,345][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:37:53,385][__main__][INFO] - Number of regex retries in iteration 476: 1 [2026-03-26 06:37:53,385][__main__][INFO] - agents played in iteration 476 are Alice, Bob [2026-03-26 06:37:54,321][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:37:54,367][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:37:54,367][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:37:54,368][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:37:54,978][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:37:55,438][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:37:55,942][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:37:56,442][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:37:56,941][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:37:57,440][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:37:57,940][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:37:58,439][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:37:58,939][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:37:59,439][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:37:59,938][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:38:00,437][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:38:00,935][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:38:01,433][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:38:01,933][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:38:02,432][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:38:02,930][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:38:03,434][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:38:03,935][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:38:04,437][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:38:04,938][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:38:05,440][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:38:05,943][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:38:06,444][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:38:06,946][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:38:07,445][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:38:07,944][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:38:08,444][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:38:08,944][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:38:09,444][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:38:09,944][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:38:10,443][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:38:10,941][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:38:11,440][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:38:11,939][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:38:12,437][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:38:12,935][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:38:13,435][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:38:13,933][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:38:14,432][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:38:14,931][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:38:15,430][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:38:15,931][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:38:16,432][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:38:16,931][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:38:17,432][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:38:17,932][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:38:18,433][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:38:18,932][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:38:19,431][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:38:19,929][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:38:20,426][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:38:20,925][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:38:21,423][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:38:21,920][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:38:22,418][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:38:22,915][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:38:23,414][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:38:23,913][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:38:24,413][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:38:24,912][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:38:25,412][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:38:25,910][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:38:26,409][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:38:26,909][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:38:27,408][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:38:27,907][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:38:28,407][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:38:28,906][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:38:29,405][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:38:29,903][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:38:30,403][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:38:30,902][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:38:31,400][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:38:31,898][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:38:32,396][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:38:32,894][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:38:33,392][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:38:33,890][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:38:34,388][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:38:34,885][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:38:35,387][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:38:35,888][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:38:36,390][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:38:36,893][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:38:37,394][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:38:37,895][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:38:38,395][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:38:38,897][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:38:39,395][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:38:39,893][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:38:40,392][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:38:40,891][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:38:41,389][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:38:41,887][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:38:42,387][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:38:42,886][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:38:43,385][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:38:43,885][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:38:44,383][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:38:44,882][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:38:45,382][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:38:45,883][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:38:46,383][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:38:46,883][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:38:47,382][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:38:47,882][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:38:48,383][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:38:48,882][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:38:49,381][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:38:49,880][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:38:50,379][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:38:50,878][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:38:51,376][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:38:51,873][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:38:52,370][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:38:52,868][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:38:53,366][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:38:53,864][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:38:54,363][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:38:54,862][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:38:55,361][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:38:55,860][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:38:56,358][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:38:56,857][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:38:57,354][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:38:57,853][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:38:58,351][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:38:58,850][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 06:38:59,487][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 06:39:00,223][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:39:00,225][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:39:00,226][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:39:00,881][__main__][INFO] - Iteration 477 took 1m 14s (9.50% Gen, 89.62% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 4m 51s. Estimated total time: 62h 9m 10s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 18s, 500 more iterations: 10h 21m 31s. [2026-03-26 06:39:00,884][__main__][INFO] - Starting iteration 477. [2026-03-26 06:39:01,286][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:39:01,286][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:39:08,046][__main__][INFO] - Number of regex retries in iteration 477: 0 [2026-03-26 06:39:08,046][__main__][INFO] - agents played in iteration 477 are Alice, Bob [2026-03-26 06:39:08,929][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:39:08,974][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:39:08,975][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:39:08,976][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:39:09,586][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:39:10,040][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:39:10,545][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:39:11,044][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:39:11,542][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:39:12,041][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:39:12,539][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:39:13,036][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:39:13,535][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:39:14,036][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:39:14,535][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:39:15,035][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:39:15,534][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:39:16,032][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:39:16,531][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:39:17,030][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:39:17,529][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:39:18,031][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:39:18,533][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:39:19,033][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:39:19,535][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:39:20,035][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:39:20,536][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:39:21,036][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:39:21,535][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:39:22,036][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:39:22,535][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:39:23,035][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:39:23,535][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:39:24,035][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:39:24,535][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:39:25,035][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:39:25,535][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:39:26,035][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:39:26,534][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:39:27,033][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:39:27,533][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:39:28,034][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:39:28,534][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:39:29,034][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:39:29,534][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:39:30,053][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:39:30,557][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:39:31,056][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:39:31,557][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:39:32,057][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:39:32,557][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:39:33,056][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:39:33,557][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:39:34,057][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:39:34,556][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:39:35,057][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:39:35,557][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:39:36,058][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:39:36,557][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:39:37,056][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:39:37,556][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:39:38,056][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:39:38,554][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:39:39,053][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:39:39,551][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:39:40,049][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:39:40,546][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:39:41,045][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:39:41,543][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:39:42,042][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:39:42,540][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:39:43,038][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:39:43,536][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:39:44,034][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:39:44,531][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:39:45,030][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:39:45,528][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:39:46,027][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:39:46,526][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:39:47,025][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:39:47,525][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:39:48,024][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:39:48,525][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:39:49,024][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:39:49,524][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:39:50,024][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:39:50,524][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:39:51,023][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:39:51,522][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:39:52,022][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:39:52,521][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:39:53,020][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:39:53,520][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:39:54,019][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:39:54,519][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:39:55,018][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:39:55,517][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:39:56,016][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:39:56,515][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:39:57,013][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:39:57,514][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:39:58,014][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:39:58,513][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:39:59,012][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:39:59,512][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:40:00,011][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:40:00,511][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:40:01,011][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:40:01,511][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:40:02,011][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:40:02,511][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:40:03,011][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:40:03,511][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:40:04,011][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:40:04,512][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:40:05,013][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:40:05,513][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:40:06,013][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:40:06,512][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:40:07,012][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:40:07,511][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:40:08,012][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:40:08,512][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:40:09,012][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:40:09,511][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:40:10,009][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:40:10,508][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:40:11,006][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:40:11,505][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:40:12,004][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:40:12,503][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:40:13,000][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:40:13,497][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:40:14,148][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 06:40:14,888][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:40:14,890][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:40:14,892][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:40:15,550][__main__][INFO] - Iteration 478 took 1m 14s (9.10% Gen, 90.01% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 47m 40s. Estimated total time: 61h 53m 14s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 46s, 500 more iterations: 10h 18m 52s. [2026-03-26 06:40:15,552][__main__][INFO] - Starting iteration 478. [2026-03-26 06:40:15,950][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:40:15,951][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:40:20,811][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:40:23,100][__main__][INFO] - Number of regex retries in iteration 478: 1 [2026-03-26 06:40:23,101][__main__][INFO] - agents played in iteration 478 are Alice, Bob [2026-03-26 06:40:23,975][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:40:24,020][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:40:24,020][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:40:24,021][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:40:24,626][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:40:25,081][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:40:25,586][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:40:26,086][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:40:26,585][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:40:27,084][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:40:27,583][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:40:28,082][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:40:28,581][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:40:29,083][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:40:29,587][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:40:30,089][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:40:30,591][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:40:31,092][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:40:31,593][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:40:32,093][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:40:32,596][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:40:33,094][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:40:33,592][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:40:34,090][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:40:34,588][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:40:35,088][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:40:35,587][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:40:36,086][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:40:36,585][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:40:37,084][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:40:37,583][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:40:38,081][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:40:38,580][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:40:39,079][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:40:39,578][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:40:40,077][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:40:40,576][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:40:41,074][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:40:41,572][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:40:42,070][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:40:42,570][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:40:43,069][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:40:43,568][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:40:44,068][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:40:44,567][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:40:45,066][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:40:45,566][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:40:46,067][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:40:46,567][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:40:47,069][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:40:47,569][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:40:48,070][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:40:48,573][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:40:49,072][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:40:49,572][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:40:50,072][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:40:50,573][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:40:51,073][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:40:51,572][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:40:52,073][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:40:52,572][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:40:53,070][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:40:53,569][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:40:54,067][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:40:54,566][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:40:55,066][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:40:55,565][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:40:56,063][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:40:56,562][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:40:57,060][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:40:57,559][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:40:58,058][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:40:58,557][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:40:59,057][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:40:59,555][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:41:00,055][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:41:00,554][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:41:01,057][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:41:01,559][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:41:02,061][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:41:02,564][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:41:03,066][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:41:03,569][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:41:04,071][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:41:04,574][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:41:05,074][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:41:05,573][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:41:06,071][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:41:06,570][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:41:07,069][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:41:07,568][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:41:08,067][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:41:08,567][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:41:09,066][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:41:09,565][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:41:10,064][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:41:10,564][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:41:11,061][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:41:11,560][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:41:12,059][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:41:12,558][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:41:13,057][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:41:13,556][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:41:14,055][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:41:14,553][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:41:15,052][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:41:15,552][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:41:16,051][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:41:16,551][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:41:17,051][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:41:17,551][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:41:18,052][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:41:18,555][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:41:19,055][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:41:19,555][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:41:20,056][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:41:20,556][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:41:21,055][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:41:21,554][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:41:22,054][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:41:22,553][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:41:23,053][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:41:23,552][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:41:24,052][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:41:24,552][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:41:25,051][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:41:25,551][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:41:26,050][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:41:26,549][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:41:27,047][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:41:27,546][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:41:28,046][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:41:28,545][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21732 tokens. [2026-03-26 06:41:29,185][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 06:41:30,028][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:41:30,030][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:41:30,032][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:41:30,755][__main__][INFO] - Iteration 479 took 1m 14s (9.56% Gen, 89.47% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 13m 28s. Estimated total time: 62h 20m 17s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 40s, 500 more iterations: 10h 23m 22s. [2026-03-26 06:41:30,757][__main__][INFO] - Starting iteration 479. [2026-03-26 06:41:31,159][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:41:31,160][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:41:34,151][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:41:34,601][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:41:38,336][__main__][INFO] - Number of regex retries in iteration 479: 2 [2026-03-26 06:41:38,336][__main__][INFO] - agents played in iteration 479 are Alice, Bob [2026-03-26 06:41:39,211][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:41:39,255][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:41:39,256][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:41:39,257][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:41:39,864][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:41:40,320][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:41:40,824][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:41:41,325][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:41:41,824][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:41:42,324][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:41:42,822][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:41:43,319][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:41:43,817][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:41:44,317][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:41:44,817][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:41:45,318][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:41:45,821][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:41:46,323][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:41:46,822][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:41:47,322][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:41:47,822][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:41:48,323][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:41:48,823][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:41:49,322][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:41:49,820][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:41:50,318][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:41:50,816][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:41:51,314][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:41:51,814][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:41:52,313][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:41:52,812][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:41:53,311][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:41:53,810][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:41:54,310][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:41:54,811][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:41:55,310][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:41:55,810][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:41:56,313][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:41:56,815][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:41:57,315][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:41:57,815][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:41:58,316][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:41:58,816][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:41:59,318][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:41:59,820][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:42:00,319][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:42:00,818][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:42:01,317][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:42:01,816][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:42:02,316][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:42:02,815][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:42:03,315][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:42:03,814][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:42:04,312][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:42:04,811][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:42:05,309][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:42:05,806][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:42:06,307][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:42:06,806][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:42:07,304][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:42:07,802][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:42:08,301][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:42:08,800][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:42:09,298][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:42:09,797][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:42:10,296][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:42:10,795][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:42:11,294][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:42:11,793][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:42:12,292][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:42:12,791][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:42:13,289][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:42:13,788][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:42:14,288][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:42:14,786][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:42:15,285][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:42:15,784][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:42:16,284][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:42:16,783][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:42:17,281][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:42:17,779][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:42:18,280][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:42:18,780][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:42:19,278][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:42:19,777][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:42:20,275][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:42:20,773][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:42:21,272][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:42:21,771][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:42:22,272][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:42:22,772][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:42:23,271][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:42:23,771][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:42:24,270][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:42:24,769][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:42:25,270][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:42:25,769][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:42:26,269][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:42:26,768][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:42:27,267][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:42:27,766][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:42:28,266][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:42:28,768][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:42:29,268][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:42:29,770][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:42:30,271][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:42:30,772][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:42:31,273][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:42:31,774][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:42:32,274][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:42:32,773][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:42:33,273][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:42:33,772][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:42:34,272][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:42:34,771][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:42:35,271][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:42:35,770][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:42:36,269][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:42:36,767][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:42:37,266][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:42:37,763][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:42:38,261][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:42:38,758][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:42:39,256][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:42:39,753][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:42:40,252][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:42:40,752][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:42:41,252][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:42:41,752][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:42:42,252][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:42:42,751][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:42:43,251][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:42:43,751][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 06:42:44,390][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 06:42:45,139][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:42:45,142][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:42:45,143][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:42:45,822][__main__][INFO] - Iteration 480 took 1m 14s (9.61% Gen, 89.48% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 5m 6s. Estimated total time: 62h 13m 9s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 26s, 500 more iterations: 10h 22m 11s. [2026-03-26 06:42:45,824][__main__][INFO] - Starting iteration 480. [2026-03-26 06:42:46,230][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:42:46,230][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:42:52,695][__main__][INFO] - Number of regex retries in iteration 480: 0 [2026-03-26 06:42:52,696][__main__][INFO] - agents played in iteration 480 are Alice, Bob [2026-03-26 06:42:53,633][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:42:53,678][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:42:53,679][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:42:53,679][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:42:54,296][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:42:54,758][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:42:55,265][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:42:55,768][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:42:56,270][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:42:56,773][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:42:57,276][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:42:57,778][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:42:58,281][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:42:58,779][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:42:59,278][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:42:59,776][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:43:00,275][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:43:00,773][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:43:01,273][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:43:01,772][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:43:02,271][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:43:02,770][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:43:03,270][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:43:03,770][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:43:04,270][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:43:04,769][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:43:05,270][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:43:05,769][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:43:06,268][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:43:06,767][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:43:07,266][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:43:07,767][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:43:08,268][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:43:08,767][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:43:09,267][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:43:09,766][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:43:10,265][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:43:10,764][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:43:11,264][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:43:11,765][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:43:12,265][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:43:12,765][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:43:13,265][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:43:13,765][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:43:14,263][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:43:14,763][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:43:15,264][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:43:15,766][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:43:16,266][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:43:16,765][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:43:17,263][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:43:17,762][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:43:18,263][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:43:18,765][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:43:19,265][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:43:19,766][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:43:20,267][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:43:20,768][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:43:21,268][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:43:21,769][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:43:22,270][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:43:22,766][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:43:23,261][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:43:23,755][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:43:24,251][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:43:24,745][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:43:25,241][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:43:25,735][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:43:26,231][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:43:26,733][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:43:27,236][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:43:27,740][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:43:28,242][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:43:28,745][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:43:29,247][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:43:29,750][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:43:30,252][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:43:30,752][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:43:31,251][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:43:31,748][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:43:32,246][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:43:32,744][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:43:33,244][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:43:33,743][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:43:34,241][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:43:34,741][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:43:35,240][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:43:35,738][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:43:36,237][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:43:36,736][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:43:37,236][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:43:37,736][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:43:38,236][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:43:38,735][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:43:39,234][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:43:39,733][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:43:40,232][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:43:40,732][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:43:41,232][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:43:41,731][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:43:42,232][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:43:42,732][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:43:43,231][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:43:43,731][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:43:44,231][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:43:44,732][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:43:45,232][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:43:45,731][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:43:46,230][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:43:46,729][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:43:47,229][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:43:47,727][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:43:48,229][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:43:48,729][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:43:49,228][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:43:49,726][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:43:50,225][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:43:50,724][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:43:51,224][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:43:51,727][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:43:52,228][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:43:52,728][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:43:53,229][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:43:53,729][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:43:54,231][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:43:54,728][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:43:55,225][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:43:55,720][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:43:56,215][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:43:56,711][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:43:57,206][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:43:57,701][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:43:58,196][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:43:58,827][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 06:43:59,566][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:43:59,568][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:43:59,570][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:44:00,295][__main__][INFO] - Iteration 481 took 1m 14s (8.73% Gen, 90.29% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 33m 59s. Estimated total time: 61h 43m 17s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 26s, 500 more iterations: 10h 17m 12s. [2026-03-26 06:44:00,297][__main__][INFO] - Starting iteration 481. [2026-03-26 06:44:00,695][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:44:00,696][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:44:08,398][__main__][INFO] - Number of regex retries in iteration 481: 0 [2026-03-26 06:44:08,399][__main__][INFO] - agents played in iteration 481 are Alice, Bob [2026-03-26 06:44:09,337][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:44:09,382][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:44:09,383][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:44:09,383][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:44:10,001][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:44:10,457][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:44:10,962][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:44:11,462][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:44:11,963][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:44:12,463][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:44:12,964][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:44:13,465][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:44:13,965][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:44:14,464][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:44:14,965][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:44:15,465][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:44:15,966][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:44:16,466][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:44:16,966][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:44:17,466][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:44:17,966][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:44:18,470][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:44:18,971][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:44:19,471][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:44:19,972][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:44:20,472][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:44:20,972][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:44:21,467][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:44:21,968][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:44:22,468][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:44:22,968][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:44:23,468][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:44:23,968][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:44:24,467][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:44:24,967][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:44:25,468][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:44:25,968][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:44:26,468][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:44:26,967][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:44:27,466][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:44:27,965][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:44:28,463][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:44:28,962][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:44:29,461][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:44:29,960][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:44:30,459][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:44:30,957][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:44:31,456][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:44:31,955][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:44:32,453][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:44:32,952][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:44:33,451][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:44:33,949][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:44:34,448][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:44:34,948][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:44:35,447][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:44:35,946][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:44:36,445][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:44:36,947][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:44:37,447][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:44:37,948][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:44:38,448][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:44:38,948][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:44:39,448][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:44:39,948][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:44:40,449][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:44:40,950][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:44:41,451][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:44:41,951][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:44:42,451][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:44:42,952][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:44:43,452][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:44:43,954][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:44:44,455][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:44:44,956][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:44:45,459][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:44:45,962][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:44:46,461][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:44:46,961][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:44:47,462][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:44:47,963][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:44:48,464][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:44:48,965][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:44:49,465][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:44:49,966][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:44:50,468][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:44:50,969][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:44:51,469][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:44:51,968][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:44:52,467][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:44:52,966][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:44:53,468][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:44:53,968][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:44:54,468][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:44:54,968][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:44:55,468][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:44:55,969][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:44:56,468][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:44:56,968][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:44:57,469][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:44:57,968][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:44:58,467][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:44:58,965][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:44:59,463][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:44:59,962][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:45:00,460][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:45:00,958][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:45:01,457][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:45:01,956][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:45:02,454][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:45:02,954][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:45:03,453][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:45:03,953][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:45:04,452][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:45:04,951][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:45:05,450][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:45:05,949][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:45:06,448][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:45:06,948][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:45:07,447][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:45:07,947][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:45:08,448][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:45:08,948][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:45:09,449][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:45:09,948][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:45:10,448][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:45:10,948][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:45:11,448][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:45:11,949][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:45:12,449][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:45:12,950][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:45:13,451][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:45:13,952][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21713 tokens. [2026-03-26 06:45:14,592][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 06:45:15,346][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:45:15,348][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:45:15,350][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:45:16,030][__main__][INFO] - Iteration 482 took 1m 15s (10.22% Gen, 88.87% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 36m 13s. Estimated total time: 62h 46m 47s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 33s, 500 more iterations: 10h 27m 47s. [2026-03-26 06:45:16,032][__main__][INFO] - Starting iteration 482. [2026-03-26 06:45:16,435][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:45:16,436][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:45:17,498][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:45:18,673][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:45:19,752][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:45:22,749][__main__][INFO] - Number of regex retries in iteration 482: 3 [2026-03-26 06:45:22,750][__main__][INFO] - agents played in iteration 482 are Alice, Bob [2026-03-26 06:45:23,635][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:45:23,962][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:45:23,963][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:45:23,963][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:45:24,582][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:45:25,039][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:45:25,544][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:45:26,044][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:45:26,545][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:45:27,045][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:45:27,544][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:45:28,045][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:45:28,546][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:45:29,047][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:45:29,547][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:45:30,047][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:45:30,547][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:45:31,045][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:45:31,546][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:45:32,046][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:45:32,546][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:45:33,045][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:45:33,544][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:45:34,044][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:45:34,543][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:45:35,042][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:45:35,542][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:45:36,041][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:45:36,541][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:45:37,041][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:45:37,541][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:45:38,039][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:45:38,538][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:45:39,038][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:45:39,537][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:45:40,037][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:45:40,536][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:45:41,035][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:45:41,534][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:45:42,034][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:45:42,534][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:45:43,034][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:45:43,534][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:45:44,034][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:45:44,534][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:45:45,034][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:45:45,533][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:45:46,034][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:45:46,534][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:45:47,033][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:45:47,532][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:45:48,031][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:45:48,532][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:45:49,034][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:45:49,535][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:45:50,036][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:45:50,537][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:45:51,037][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:45:51,538][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:45:52,038][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:45:52,538][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:45:53,038][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:45:53,538][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:45:54,038][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:45:54,536][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:45:55,037][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:45:55,536][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:45:56,035][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:45:56,535][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:45:57,036][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:45:57,537][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:45:58,037][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:45:58,538][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:45:59,037][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:45:59,538][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:46:00,038][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:46:00,539][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:46:01,039][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:46:01,538][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:46:02,038][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:46:02,538][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:46:03,037][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:46:03,537][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:46:04,036][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:46:04,536][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:46:05,036][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:46:05,536][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:46:06,034][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:46:06,534][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:46:07,034][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:46:07,534][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:46:08,034][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:46:08,534][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:46:09,034][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:46:09,534][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:46:10,034][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:46:10,533][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:46:11,033][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:46:11,534][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:46:12,034][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:46:12,534][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:46:13,033][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:46:13,533][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:46:14,033][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:46:14,533][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:46:15,033][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:46:15,536][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:46:16,039][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:46:16,538][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:46:17,036][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:46:17,536][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:46:18,036][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:46:18,538][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:46:19,038][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:46:19,538][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:46:20,036][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:46:20,535][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:46:21,036][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:46:21,538][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:46:22,039][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:46:22,540][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:46:23,040][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:46:23,541][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:46:24,042][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:46:24,542][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:46:25,042][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:46:25,542][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:46:26,042][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:46:26,542][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:46:27,041][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:46:27,540][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:46:28,039][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:46:28,539][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 06:46:29,183][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 06:46:29,932][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:46:29,934][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:46:29,935][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:46:30,597][__main__][INFO] - Iteration 483 took 1m 14s (8.51% Gen, 90.59% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 51h 36m 19s. Estimated total time: 61h 48m 7s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 36s, 500 more iterations: 10h 18m 1s. [2026-03-26 06:46:30,599][__main__][INFO] - Starting iteration 483. [2026-03-26 06:46:31,001][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:46:31,002][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:46:32,108][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 balls, 10 books did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:46:32,204][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:46:37,710][__main__][INFO] - Number of regex retries in iteration 483: 2 [2026-03-26 06:46:37,711][__main__][INFO] - agents played in iteration 483 are Alice, Bob [2026-03-26 06:46:38,587][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:46:38,632][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:46:38,633][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:46:38,633][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:46:39,242][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:46:39,697][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:46:40,200][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:46:40,699][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:46:41,199][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:46:41,697][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:46:42,197][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:46:42,695][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:46:43,195][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:46:43,695][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:46:44,194][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:46:44,693][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:46:45,193][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:46:45,693][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:46:46,194][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:46:46,694][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:46:47,194][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:46:47,692][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:46:48,191][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:46:48,692][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:46:49,191][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:46:49,690][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:46:50,189][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:46:50,688][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:46:51,187][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:46:51,686][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:46:52,185][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:46:52,684][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:46:53,182][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:46:53,682][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:46:54,181][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:46:54,680][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:46:55,180][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:46:55,679][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:46:56,178][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:46:56,676][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:46:57,175][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:46:57,675][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:46:58,173][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:46:58,673][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:46:59,172][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:46:59,670][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:47:00,169][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:47:00,668][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:47:01,167][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:47:01,666][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:47:02,164][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:47:02,662][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:47:03,160][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:47:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:47:04,162][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:47:04,662][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:47:05,161][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:47:05,661][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:47:06,160][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:47:06,661][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:47:07,160][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:47:07,659][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:47:08,158][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:47:08,657][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:47:09,156][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:47:09,655][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:47:10,154][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:47:10,653][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:47:11,152][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:47:11,652][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:47:12,151][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:47:12,651][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:47:13,152][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:47:13,651][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:47:14,151][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:47:14,650][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:47:15,150][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:47:15,649][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:47:16,147][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:47:16,647][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:47:17,146][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:47:17,645][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:47:18,144][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:47:18,644][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:47:19,144][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:47:19,642][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:47:20,141][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:47:20,640][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:47:21,140][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:47:21,639][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:47:22,138][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:47:22,636][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:47:23,136][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:47:23,636][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:47:24,134][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:47:24,633][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:47:25,133][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:47:25,632][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:47:26,131][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:47:26,630][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:47:27,130][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:47:27,629][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:47:28,128][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:47:28,626][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:47:29,125][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:47:29,624][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:47:30,123][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:47:30,621][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:47:31,119][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:47:31,618][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:47:32,115][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:47:32,614][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:47:33,113][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:47:33,611][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:47:34,109][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:47:34,610][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:47:35,109][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:47:35,609][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:47:36,107][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:47:36,606][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:47:37,106][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:47:37,605][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:47:38,106][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:47:38,607][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:47:39,107][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:47:39,606][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:47:40,105][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:47:40,603][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:47:41,102][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:47:41,600][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:47:42,098][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:47:42,597][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:47:43,097][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 06:47:43,735][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 06:47:44,545][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:47:44,548][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:47:44,549][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:47:45,563][__main__][INFO] - Iteration 484 took 1m 14s (9.00% Gen, 89.64% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 55m 3s. Estimated total time: 62h 8m 7s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 16s, 500 more iterations: 10h 21m 21s. [2026-03-26 06:47:45,565][__main__][INFO] - Starting iteration 484. [2026-03-26 06:47:45,963][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:47:45,964][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:47:53,401][__main__][INFO] - Number of regex retries in iteration 484: 0 [2026-03-26 06:47:53,402][__main__][INFO] - agents played in iteration 484 are Alice, Bob [2026-03-26 06:47:54,289][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:47:54,334][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:47:54,334][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:47:54,335][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:47:54,944][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:47:55,400][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:47:55,904][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:47:56,403][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:47:56,902][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:47:57,402][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:47:57,901][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:47:58,400][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:47:58,900][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:47:59,398][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:47:59,898][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:48:00,396][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:48:00,895][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:48:01,394][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:48:01,892][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:48:02,391][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:48:02,889][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:48:03,388][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:48:03,887][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:48:04,386][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:48:04,884][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:48:05,382][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:48:05,880][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:48:06,378][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:48:06,877][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:48:07,379][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:48:07,880][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:48:08,380][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:48:08,881][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:48:09,380][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:48:09,880][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:48:10,380][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:48:10,881][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:48:11,381][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:48:11,882][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:48:12,381][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:48:12,881][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:48:13,381][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:48:13,881][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:48:14,380][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:48:14,880][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:48:15,379][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:48:15,878][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:48:16,378][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:48:16,878][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:48:17,377][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:48:17,877][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:48:18,378][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:48:18,878][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:48:19,376][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:48:19,874][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:48:20,372][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:48:20,871][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:48:21,370][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:48:21,869][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:48:22,368][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:48:22,867][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:48:23,365][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:48:23,862][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:48:24,360][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:48:24,858][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:48:25,357][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:48:25,856][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:48:26,354][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:48:26,853][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:48:27,352][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:48:27,851][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:48:28,351][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:48:28,849][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:48:29,350][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:48:29,849][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:48:30,349][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:48:30,849][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:48:31,348][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:48:31,847][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:48:32,345][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:48:32,844][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:48:33,344][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:48:33,843][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:48:34,342][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:48:34,841][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:48:35,340][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:48:35,838][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:48:36,337][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:48:36,836][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:48:37,335][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:48:37,834][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:48:38,332][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:48:38,831][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:48:39,331][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:48:39,832][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:48:40,333][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:48:40,833][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:48:41,333][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:48:41,834][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:48:42,335][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:48:42,835][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:48:43,335][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:48:43,835][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:48:44,335][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:48:44,834][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:48:45,335][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:48:45,835][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:48:46,335][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:48:46,835][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:48:47,335][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:48:47,835][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:48:48,335][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:48:48,834][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:48:49,333][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:48:49,833][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:48:50,332][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:48:50,831][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:48:51,330][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:48:51,828][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:48:52,327][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:48:52,825][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:48:53,322][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:48:53,820][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:48:54,317][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:48:54,815][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:48:55,315][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:48:55,814][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:48:56,312][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:48:56,811][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:48:57,309][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:48:57,807][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:48:58,306][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:48:58,804][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 06:48:59,441][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 06:49:00,175][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:49:00,177][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:49:00,178][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:49:01,031][__main__][INFO] - Iteration 485 took 1m 15s (9.91% Gen, 88.95% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 19m 5s. Estimated total time: 62h 33m 24s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 6s, 500 more iterations: 10h 25m 34s. [2026-03-26 06:49:01,033][__main__][INFO] - Starting iteration 485. [2026-03-26 06:49:01,433][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:49:01,434][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:49:03,174][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:49:08,332][__main__][INFO] - Number of regex retries in iteration 485: 1 [2026-03-26 06:49:08,333][__main__][INFO] - agents played in iteration 485 are Alice, Bob [2026-03-26 06:49:09,221][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:49:09,266][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:49:09,266][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:49:09,267][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:49:09,874][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:49:10,330][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:49:10,834][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:49:11,334][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:49:11,834][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:49:12,335][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:49:12,834][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:49:13,334][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:49:13,834][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:49:14,335][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:49:14,835][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:49:15,336][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:49:15,840][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:49:16,340][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:49:16,841][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:49:17,340][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:49:17,840][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:49:18,343][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:49:18,844][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:49:19,344][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:49:19,845][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:49:20,345][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:49:20,844][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:49:21,343][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:49:21,843][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:49:22,342][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:49:22,841][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:49:23,341][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:49:23,841][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:49:24,341][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:49:24,840][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:49:25,340][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:49:25,840][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:49:26,338][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:49:26,838][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:49:27,338][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:49:27,836][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:49:28,335][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:49:28,835][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:49:29,334][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:49:29,833][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:49:30,331][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:49:30,830][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:49:31,328][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:49:31,826][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:49:32,324][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:49:32,822][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:49:33,320][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:49:33,818][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:49:34,319][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:49:34,819][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:49:35,318][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:49:35,818][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:49:36,317][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:49:36,818][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:49:37,318][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:49:37,819][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:49:38,320][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:49:38,819][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:49:39,318][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:49:39,818][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:49:40,317][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:49:40,817][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:49:41,318][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:49:41,818][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:49:42,318][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:49:42,817][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:49:43,315][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:49:43,814][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:49:44,313][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:49:44,813][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:49:45,312][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:49:45,812][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:49:46,311][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:49:46,812][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:49:47,312][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:49:47,811][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:49:48,315][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:49:48,816][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:49:49,317][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:49:49,818][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:49:50,318][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:49:50,818][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:49:51,318][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:49:51,819][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:49:52,320][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:49:52,819][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:49:53,319][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:49:53,818][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:49:54,318][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:49:54,817][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:49:55,319][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:49:55,819][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:49:56,318][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:49:56,818][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:49:57,318][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:49:57,817][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:49:58,318][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:49:58,818][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:49:59,318][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:49:59,818][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:50:00,318][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:50:00,817][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:50:01,317][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:50:01,816][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:50:02,316][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:50:02,815][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:50:03,314][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:50:03,812][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:50:04,311][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:50:04,811][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:50:05,309][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:50:05,808][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:50:06,310][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:50:06,810][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:50:07,309][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:50:07,809][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:50:08,311][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:50:08,811][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:50:09,311][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:50:09,811][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:50:10,312][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:50:10,812][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:50:11,312][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:50:11,813][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:50:12,314][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:50:12,814][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:50:13,314][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:50:13,814][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 06:50:14,453][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 06:50:15,190][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:50:15,193][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:50:15,194][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:50:15,843][__main__][INFO] - Iteration 486 took 1m 14s (9.27% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 44m 57s. Estimated total time: 62h 0m 31s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 1s, 500 more iterations: 10h 20m 5s. [2026-03-26 06:50:15,845][__main__][INFO] - Starting iteration 486. [2026-03-26 06:50:16,248][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:50:16,249][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:50:23,464][__main__][INFO] - Number of regex retries in iteration 486: 0 [2026-03-26 06:50:23,464][__main__][INFO] - agents played in iteration 486 are Alice, Bob [2026-03-26 06:50:24,353][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:50:24,398][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:50:24,398][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:50:24,399][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:50:25,004][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:50:25,462][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:50:25,966][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:50:26,466][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:50:26,965][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:50:27,464][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:50:27,964][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:50:28,464][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:50:28,963][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:50:29,461][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:50:29,960][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:50:30,458][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:50:30,957][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:50:31,456][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:50:31,955][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:50:32,453][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:50:32,952][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:50:33,452][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:50:33,952][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:50:34,452][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:50:34,951][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:50:35,451][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:50:35,950][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:50:36,449][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:50:36,948][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:50:37,452][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:50:37,954][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:50:38,454][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:50:38,956][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:50:39,456][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:50:39,958][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:50:40,459][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:50:40,959][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:50:41,458][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:50:41,958][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:50:42,457][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:50:42,956][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:50:43,455][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:50:43,955][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:50:44,455][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:50:44,954][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:50:45,457][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:50:45,959][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:50:46,458][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:50:46,957][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:50:47,456][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:50:47,957][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:50:48,459][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:50:48,958][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:50:49,458][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:50:49,956][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:50:50,458][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:50:50,958][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:50:51,456][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:50:51,956][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:50:52,455][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:50:52,954][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:50:53,453][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:50:53,953][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:50:54,453][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:50:54,953][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:50:55,453][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:50:55,953][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:50:56,452][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:50:56,952][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:50:57,451][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:50:57,950][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:50:58,451][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:50:58,951][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:50:59,451][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:50:59,950][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:51:00,448][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:51:00,948][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:51:01,446][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:51:01,944][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:51:02,442][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:51:02,941][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:51:03,440][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:51:03,938][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:51:04,437][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:51:04,936][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:51:05,436][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:51:05,936][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:51:06,436][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:51:06,937][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:51:07,436][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:51:07,936][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:51:08,436][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:51:08,936][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:51:09,436][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:51:09,937][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:51:10,438][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:51:10,939][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:51:11,440][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:51:11,941][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:51:12,442][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:51:12,943][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:51:13,442][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:51:13,941][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:51:14,441][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:51:14,940][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:51:15,439][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:51:15,938][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:51:16,438][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:51:16,938][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:51:17,437][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:51:17,937][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:51:18,439][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:51:18,938][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:51:19,437][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:51:19,936][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:51:20,437][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:51:20,938][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:51:21,437][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:51:21,936][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:51:22,435][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:51:22,934][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:51:23,434][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:51:23,933][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:51:24,432][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:51:24,931][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:51:25,432][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:51:25,931][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:51:26,432][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:51:26,932][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:51:27,431][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:51:27,931][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:51:28,431][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:51:28,930][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 06:51:29,559][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 06:51:30,293][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:51:30,295][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:51:30,297][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:51:30,991][__main__][INFO] - Iteration 487 took 1m 14s (9.65% Gen, 89.41% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 0m 23s. Estimated total time: 62h 17m 12s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 34s, 500 more iterations: 10h 22m 52s. [2026-03-26 06:51:30,994][__main__][INFO] - Starting iteration 487. [2026-03-26 06:51:31,397][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:51:31,398][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:51:37,833][mllm.models.large_language_model_local][WARNING] - Response Given that both Alice and Bob have similar values for hats and books, and balls, it's a balanced scenario where proposing to split the items evenly could maximize the points for both parties, aligning with the previous rounds' outcomes. Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:51:39,035][__main__][INFO] - Number of regex retries in iteration 487: 1 [2026-03-26 06:51:39,035][__main__][INFO] - agents played in iteration 487 are Alice, Bob [2026-03-26 06:51:39,971][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:51:40,016][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:51:40,017][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:51:40,018][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:51:40,632][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:51:41,089][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:51:41,593][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:51:42,092][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:51:42,592][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:51:43,091][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:51:43,589][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:51:44,090][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:51:44,590][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:51:45,090][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:51:45,590][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:51:46,089][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:51:46,588][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:51:47,087][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:51:47,588][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:51:48,089][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:51:48,590][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:51:49,092][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:51:49,593][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:51:50,093][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:51:50,595][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:51:51,095][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:51:51,595][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:51:52,098][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:51:52,598][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:51:53,097][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:51:53,596][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:51:54,097][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:51:54,597][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:51:55,097][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:51:55,597][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:51:56,097][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:51:56,597][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:51:57,100][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:51:57,617][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:51:58,118][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:51:58,618][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:51:59,118][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:51:59,619][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:52:00,121][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:52:00,621][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:52:01,122][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:52:01,621][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:52:02,121][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:52:02,619][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:52:03,119][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:52:03,619][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:52:04,120][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:52:04,620][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:52:05,119][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:52:05,618][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:52:06,117][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:52:06,617][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:52:07,116][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:52:07,615][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:52:08,115][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:52:08,614][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:52:09,113][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:52:09,613][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:52:10,112][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:52:10,612][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:52:11,112][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:52:11,611][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:52:12,111][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:52:12,611][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:52:13,110][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:52:13,610][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:52:14,109][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:52:14,610][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:52:15,111][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:52:15,614][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:52:16,115][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:52:16,615][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:52:17,115][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:52:17,616][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:52:18,117][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:52:18,618][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:52:19,118][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:52:19,617][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:52:20,117][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:52:20,617][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:52:21,118][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:52:21,618][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:52:22,118][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:52:22,617][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:52:23,117][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:52:23,617][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:52:24,117][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:52:24,618][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:52:25,119][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:52:25,619][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:52:26,119][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:52:26,618][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:52:27,117][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:52:27,617][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:52:28,117][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:52:28,617][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:52:29,118][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:52:29,618][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:52:30,118][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:52:30,618][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:52:31,119][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:52:31,619][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:52:32,120][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:52:32,619][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:52:33,118][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:52:33,618][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:52:34,117][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:52:34,615][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:52:35,115][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:52:35,614][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:52:36,114][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:52:36,614][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:52:37,113][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:52:37,613][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:52:38,112][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:52:38,611][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:52:39,112][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:52:39,612][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:52:40,111][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:52:40,609][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:52:41,107][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:52:41,605][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:52:42,103][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:52:42,601][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:52:43,100][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:52:43,600][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:52:44,099][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:52:44,598][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 06:52:45,236][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 06:52:45,980][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:52:45,985][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:52:45,988][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:52:46,704][__main__][INFO] - Iteration 488 took 1m 15s (10.14% Gen, 88.91% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 27m 17s. Estimated total time: 62h 45m 22s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 30s, 500 more iterations: 10h 27m 33s. [2026-03-26 06:52:46,706][__main__][INFO] - Starting iteration 488. [2026-03-26 06:52:47,111][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:52:47,112][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:52:52,260][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:52:53,448][__main__][INFO] - Number of regex retries in iteration 488: 1 [2026-03-26 06:52:53,449][__main__][INFO] - agents played in iteration 488 are Alice, Bob [2026-03-26 06:52:54,324][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:52:54,654][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:52:54,655][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:52:54,655][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:52:55,264][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:52:55,719][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:52:56,223][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:52:56,722][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:52:57,221][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:52:57,720][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:52:58,219][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:52:58,719][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:52:59,217][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:52:59,717][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:53:00,217][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:53:00,716][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:53:01,216][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:53:01,715][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:53:02,215][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:53:02,714][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:53:03,214][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:53:03,713][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:53:04,215][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:53:04,716][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:53:05,216][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:53:05,717][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:53:06,216][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:53:06,716][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:53:07,217][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:53:07,717][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:53:08,217][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:53:08,718][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:53:09,219][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:53:09,720][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:53:10,220][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:53:10,720][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:53:11,219][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:53:11,718][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:53:12,219][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:53:12,719][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:53:13,218][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:53:13,717][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:53:14,217][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:53:14,716][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:53:15,215][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:53:15,714][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:53:16,215][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:53:16,715][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:53:17,215][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:53:17,715][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:53:18,216][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:53:18,719][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:53:19,220][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:53:19,721][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:53:20,221][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:53:20,720][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:53:21,220][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:53:21,719][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:53:22,219][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:53:22,719][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:53:23,220][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:53:23,720][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:53:24,219][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:53:24,719][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:53:25,219][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:53:25,718][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:53:26,217][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:53:26,717][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:53:27,216][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:53:27,716][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:53:28,215][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:53:28,714][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:53:29,214][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:53:29,713][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:53:30,213][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:53:30,712][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:53:31,211][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:53:31,711][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:53:32,210][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:53:32,711][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:53:33,211][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:53:33,712][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:53:34,212][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:53:34,712][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:53:35,211][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:53:35,711][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:53:36,211][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:53:36,712][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:53:37,212][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:53:37,713][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:53:38,213][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:53:38,713][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:53:39,213][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:53:39,714][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:53:40,214][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:53:40,714][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:53:41,214][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:53:41,715][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:53:42,215][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:53:42,715][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:53:43,215][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:53:43,715][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:53:44,215][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:53:44,714][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:53:45,214][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:53:45,715][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:53:46,217][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:53:46,716][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:53:47,215][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:53:47,716][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:53:48,220][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:53:48,720][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:53:49,221][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:53:49,720][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:53:50,220][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:53:50,720][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:53:51,221][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:53:51,721][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:53:52,222][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:53:52,722][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:53:53,222][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:53:53,723][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:53:54,223][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:53:54,723][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:53:55,224][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:53:55,724][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:53:56,223][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:53:56,721][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:53:57,220][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:53:57,719][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:53:58,219][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:53:58,718][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:53:59,218][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21713 tokens. [2026-03-26 06:53:59,854][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 06:54:00,592][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:54:00,594][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:54:00,596][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:54:01,261][__main__][INFO] - Iteration 489 took 1m 14s (8.55% Gen, 90.55% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 51h 28m 13s. Estimated total time: 61h 47m 32s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 55s. [2026-03-26 06:54:01,263][__main__][INFO] - Starting iteration 489. [2026-03-26 06:54:01,669][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:54:01,670][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:54:02,253][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:54:07,435][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:54:07,465][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:54:08,508][__main__][INFO] - Number of regex retries in iteration 489: 3 [2026-03-26 06:54:08,509][__main__][INFO] - agents played in iteration 489 are Alice, Bob [2026-03-26 06:54:09,446][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:54:09,491][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:54:09,492][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:54:09,493][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:54:10,109][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:54:10,566][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:54:11,070][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:54:11,568][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:54:12,066][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:54:12,565][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:54:13,065][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:54:13,565][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:54:14,064][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:54:14,563][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:54:15,063][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:54:15,562][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:54:16,061][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:54:16,561][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:54:17,063][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:54:17,563][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:54:18,063][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:54:18,563][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:54:19,064][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:54:19,563][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:54:20,062][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:54:20,562][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:54:21,062][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:54:21,561][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:54:22,061][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:54:22,560][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:54:23,059][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:54:23,559][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:54:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:54:24,556][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:54:25,056][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:54:25,555][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:54:26,053][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:54:26,553][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:54:27,054][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:54:27,554][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:54:28,054][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:54:28,554][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:54:29,054][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:54:29,554][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:54:30,053][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:54:30,555][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:54:31,056][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:54:31,556][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:54:32,056][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:54:32,556][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:54:33,057][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:54:33,557][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:54:34,057][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:54:34,556][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:54:35,056][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:54:35,555][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:54:36,054][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:54:36,553][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:54:37,052][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:54:37,553][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:54:38,052][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:54:38,553][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:54:39,053][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:54:39,552][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:54:40,052][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:54:40,552][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:54:41,053][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:54:41,553][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:54:42,053][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:54:42,552][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:54:43,051][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:54:43,550][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:54:44,048][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:54:44,548][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:54:45,047][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:54:45,546][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:54:46,046][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:54:46,545][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:54:47,044][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:54:47,542][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:54:48,041][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:54:48,543][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:54:49,043][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:54:49,542][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:54:50,041][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:54:50,539][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:54:51,039][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:54:51,539][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:54:52,038][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:54:52,537][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:54:53,037][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:54:53,537][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:54:54,037][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:54:54,536][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:54:55,037][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:54:55,537][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:54:56,035][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:54:56,533][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:54:57,031][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:54:57,531][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:54:58,028][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:54:58,527][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:54:59,026][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:54:59,525][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:55:00,024][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:55:00,523][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:55:01,021][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:55:01,520][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:55:02,019][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:55:02,518][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:55:03,019][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:55:03,519][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:55:04,018][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:55:04,519][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:55:05,019][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:55:05,519][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:55:06,018][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:55:06,517][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:55:07,017][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:55:07,517][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:55:08,016][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:55:08,515][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:55:09,014][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:55:09,513][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:55:10,012][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:55:10,513][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:55:11,013][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:55:11,513][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:55:12,013][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:55:12,513][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:55:13,013][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:55:13,513][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:55:14,013][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 06:55:14,647][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 06:55:15,388][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:55:15,391][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:55:15,392][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:55:16,039][__main__][INFO] - Iteration 490 took 1m 14s (9.20% Gen, 89.93% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 37m 58s. Estimated total time: 61h 58m 32s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 57s, 500 more iterations: 10h 19m 45s. [2026-03-26 06:55:16,042][__main__][INFO] - Starting iteration 490. [2026-03-26 06:55:16,442][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:55:16,443][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:55:23,659][__main__][INFO] - Number of regex retries in iteration 490: 0 [2026-03-26 06:55:23,660][__main__][INFO] - agents played in iteration 490 are Alice, Bob [2026-03-26 06:55:24,537][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:55:24,582][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:55:24,583][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:55:24,584][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:55:25,194][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:55:25,650][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:55:26,154][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:55:26,654][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:55:27,153][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:55:27,653][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:55:28,152][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:55:28,652][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:55:29,153][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:55:29,654][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:55:30,156][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:55:30,657][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:55:31,158][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:55:31,658][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:55:32,159][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:55:32,660][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:55:33,161][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:55:33,660][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:55:34,158][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:55:34,659][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:55:35,158][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:55:35,657][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:55:36,156][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:55:36,656][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:55:37,155][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:55:37,652][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:55:38,150][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:55:38,649][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:55:39,146][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:55:39,643][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:55:40,140][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:55:40,638][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:55:41,135][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:55:41,634][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:55:42,134][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:55:42,635][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:55:43,134][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:55:43,634][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:55:44,134][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:55:44,634][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:55:45,134][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:55:45,634][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:55:46,136][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:55:46,636][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:55:47,136][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:55:47,636][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:55:48,138][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:55:48,640][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:55:49,139][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:55:49,639][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:55:50,138][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:55:50,637][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:55:51,136][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:55:51,635][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:55:52,134][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:55:52,634][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:55:53,134][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:55:53,633][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:55:54,132][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:55:54,632][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:55:55,133][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:55:55,632][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:55:56,133][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:55:56,633][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:55:57,133][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:55:57,633][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:55:58,133][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:55:58,633][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:55:59,133][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:55:59,632][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:56:00,132][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:56:00,634][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:56:01,134][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:56:01,635][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:56:02,135][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:56:02,636][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:56:03,137][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:56:03,637][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:56:04,138][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:56:04,638][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:56:05,140][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:56:05,639][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:56:06,137][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:56:06,638][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:56:07,137][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:56:07,637][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:56:08,135][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:56:08,635][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:56:09,134][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:56:09,632][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:56:10,129][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:56:10,626][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:56:11,123][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:56:11,620][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:56:12,117][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:56:12,615][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:56:13,113][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:56:13,612][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:56:14,111][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:56:14,611][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:56:15,110][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:56:15,610][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:56:16,110][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:56:16,609][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:56:17,109][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:56:17,609][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:56:18,110][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:56:18,611][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:56:19,112][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:56:19,614][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:56:20,114][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:56:20,615][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:56:21,115][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:56:21,615][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:56:22,115][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:56:22,614][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:56:23,113][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:56:23,613][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:56:24,112][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:56:24,612][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:56:25,111][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:56:25,610][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:56:26,108][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:56:26,606][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:56:27,105][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:56:27,604][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:56:28,103][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:56:28,601][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:56:29,100][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 06:56:29,735][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 06:56:30,472][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:56:30,476][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:56:30,478][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:56:31,313][__main__][INFO] - Iteration 491 took 1m 14s (9.64% Gen, 89.24% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 1m 44s. Estimated total time: 62h 23m 34s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 47s, 500 more iterations: 10h 23m 55s. [2026-03-26 06:56:31,317][__main__][INFO] - Starting iteration 491. [2026-03-26 06:56:31,718][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:56:31,718][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:56:38,219][__main__][INFO] - Number of regex retries in iteration 491: 0 [2026-03-26 06:56:38,219][__main__][INFO] - agents played in iteration 491 are Alice, Bob [2026-03-26 06:56:39,096][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:56:39,141][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:56:39,141][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:56:39,142][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:56:39,747][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:56:40,202][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:56:40,705][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:56:41,204][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:56:41,703][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:56:42,200][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:56:42,698][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:56:43,197][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:56:43,695][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:56:44,192][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:56:44,687][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:56:45,183][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:56:45,682][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:56:46,179][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:56:46,675][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:56:47,170][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:56:47,665][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:56:48,169][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:56:48,671][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:56:49,172][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:56:49,673][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:56:50,173][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:56:50,673][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:56:51,173][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:56:51,675][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:56:52,174][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:56:52,673][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:56:53,172][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:56:53,671][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:56:54,170][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:56:54,669][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:56:55,168][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:56:55,667][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:56:56,167][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:56:56,667][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:56:57,166][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:56:57,664][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:56:58,162][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:56:58,661][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:56:59,159][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:56:59,659][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:57:00,158][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:57:00,659][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:57:01,158][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:57:01,657][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:57:02,156][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:57:02,656][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:57:03,155][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:57:03,655][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:57:04,155][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:57:04,655][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:57:05,153][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:57:05,653][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:57:06,153][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:57:06,653][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:57:07,153][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:57:07,653][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:57:08,154][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:57:08,654][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:57:09,153][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:57:09,653][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:57:10,154][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:57:10,654][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:57:11,154][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:57:11,654][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:57:12,152][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:57:12,650][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:57:13,147][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:57:13,645][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:57:14,144][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:57:14,642][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:57:15,141][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:57:15,639][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:57:16,135][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:57:16,630][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:57:17,124][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:57:17,619][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:57:18,116][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:57:18,614][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:57:19,110][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:57:19,606][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:57:20,105][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:57:20,606][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:57:21,108][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:57:21,608][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:57:22,110][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:57:22,612][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:57:23,113][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:57:23,614][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:57:24,113][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:57:24,611][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:57:25,111][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:57:25,610][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:57:26,110][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:57:26,610][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:57:27,109][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:57:27,608][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:57:28,105][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:57:28,603][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:57:29,101][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:57:29,600][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:57:30,099][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:57:30,598][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:57:31,097][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:57:31,596][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:57:32,095][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:57:32,595][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:57:33,093][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:57:33,593][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:57:34,094][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:57:34,594][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:57:35,093][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:57:35,593][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:57:36,092][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:57:36,590][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:57:37,089][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:57:37,588][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:57:38,086][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:57:38,585][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:57:39,084][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:57:39,583][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:57:40,083][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:57:40,582][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:57:41,081][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:57:41,582][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:57:42,081][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:57:42,582][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:57:43,082][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:57:43,582][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 06:57:44,211][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.38%, ΔTime: 00:01:04 [2026-03-26 06:57:44,953][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:57:44,956][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:57:44,958][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:57:45,776][__main__][INFO] - Iteration 492 took 1m 14s (8.78% Gen, 90.12% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 19m 51s. Estimated total time: 61h 42m 55s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 25s, 500 more iterations: 10h 17m 9s. [2026-03-26 06:57:45,778][__main__][INFO] - Starting iteration 492. [2026-03-26 06:57:46,183][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:57:46,183][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:57:46,773][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 06:57:53,090][__main__][INFO] - Number of regex retries in iteration 492: 1 [2026-03-26 06:57:53,091][__main__][INFO] - agents played in iteration 492 are Alice, Bob [2026-03-26 06:57:54,031][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:57:54,355][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:57:54,356][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:57:54,356][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:57:54,975][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:57:55,431][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:57:55,936][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:57:56,437][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:57:56,938][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:57:57,439][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:57:57,941][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:57:58,442][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:57:58,942][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:57:59,443][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:57:59,945][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:58:00,445][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:58:00,945][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:58:01,446][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:58:01,946][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:58:02,447][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:58:02,948][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:58:03,447][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:58:03,947][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:58:04,446][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:58:04,945][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:58:05,443][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:58:05,943][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:58:06,442][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:58:06,942][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:58:07,443][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:58:07,944][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:58:08,444][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:58:08,943][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:58:09,443][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:58:09,944][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:58:10,445][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:58:10,945][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:58:11,445][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:58:11,944][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:58:12,443][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:58:12,943][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:58:13,442][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:58:13,941][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:58:14,441][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:58:14,940][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:58:15,442][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:58:15,944][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:58:16,444][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:58:16,943][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:58:17,443][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:58:17,942][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:58:18,444][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:58:18,942][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:58:19,441][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:58:19,941][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:58:20,441][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:58:20,939][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:58:21,439][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:58:21,938][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:58:22,438][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:58:22,937][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:58:23,437][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:58:23,938][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:58:24,438][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:58:24,939][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:58:25,440][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:58:25,941][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:58:26,440][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:58:26,941][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:58:27,442][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:58:27,943][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:58:28,443][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:58:28,943][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:58:29,442][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:58:29,942][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:58:30,443][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:58:30,943][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:58:31,444][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:58:31,944][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:58:32,445][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:58:32,945][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:58:33,446][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:58:33,947][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:58:34,448][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:58:34,949][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:58:35,448][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:58:35,946][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:58:36,445][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:58:36,944][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:58:37,443][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:58:37,943][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:58:38,442][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:58:38,942][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:58:39,441][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:58:39,941][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:58:40,441][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:58:40,941][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:58:41,442][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:58:41,942][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:58:42,443][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:58:42,942][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:58:43,441][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:58:43,941][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:58:44,439][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:58:44,940][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:58:45,440][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 06:58:45,939][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 06:58:46,439][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 06:58:46,938][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 06:58:47,437][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 06:58:47,938][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 06:58:48,438][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 06:58:48,938][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 06:58:49,437][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 06:58:49,935][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 06:58:50,434][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 06:58:50,935][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 06:58:51,434][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 06:58:51,934][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 06:58:52,434][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 06:58:52,934][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 06:58:53,433][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 06:58:53,931][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 06:58:54,430][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 06:58:54,928][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 06:58:55,428][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 06:58:55,928][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 06:58:56,427][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 06:58:56,926][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 06:58:57,426][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 06:58:57,926][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 06:58:58,425][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 06:58:58,925][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 06:58:59,567][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 06:59:00,308][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 06:59:00,311][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 06:59:00,312][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 06:59:00,974][__main__][INFO] - Iteration 493 took 1m 14s (9.24% Gen, 89.88% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 51h 55m 21s. Estimated total time: 62h 19m 40s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 39s, 500 more iterations: 10h 23m 16s. [2026-03-26 06:59:00,977][__main__][INFO] - Starting iteration 493. [2026-03-26 06:59:01,374][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 06:59:01,375][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 06:59:07,905][__main__][INFO] - Number of regex retries in iteration 493: 0 [2026-03-26 06:59:07,906][__main__][INFO] - agents played in iteration 493 are Alice, Bob [2026-03-26 06:59:08,843][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:59:08,888][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 06:59:08,889][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 06:59:08,890][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 06:59:09,503][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 06:59:09,959][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 06:59:10,463][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 06:59:10,962][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 06:59:11,462][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 06:59:11,961][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 06:59:12,461][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 06:59:12,961][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 06:59:13,460][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 06:59:13,959][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 06:59:14,459][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 06:59:14,959][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 06:59:15,459][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 06:59:15,959][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 06:59:16,459][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 06:59:16,959][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 06:59:17,459][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 06:59:17,960][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 06:59:18,461][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 06:59:18,961][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 06:59:19,459][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 06:59:19,959][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 06:59:20,457][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 06:59:20,956][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 06:59:21,455][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 06:59:21,956][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 06:59:22,457][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 06:59:22,957][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 06:59:23,458][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 06:59:23,959][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 06:59:24,459][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 06:59:24,960][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 06:59:25,461][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 06:59:25,962][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 06:59:26,460][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 06:59:26,959][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 06:59:27,459][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 06:59:27,958][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 06:59:28,456][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 06:59:28,955][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 06:59:29,455][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 06:59:29,955][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 06:59:30,455][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 06:59:30,954][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 06:59:31,454][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 06:59:31,954][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 06:59:32,454][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 06:59:32,955][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 06:59:33,455][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 06:59:33,955][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 06:59:34,455][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 06:59:34,954][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 06:59:35,454][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 06:59:35,953][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 06:59:36,455][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 06:59:36,955][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 06:59:37,455][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 06:59:37,955][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 06:59:38,455][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 06:59:38,954][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 06:59:39,453][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 06:59:39,954][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 06:59:40,454][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 06:59:40,954][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 06:59:41,454][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 06:59:41,954][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 06:59:42,453][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 06:59:42,953][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 06:59:43,454][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 06:59:43,954][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 06:59:44,454][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 06:59:44,954][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 06:59:45,457][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 06:59:45,959][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 06:59:46,459][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 06:59:46,960][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 06:59:47,460][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 06:59:47,960][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 06:59:48,462][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 06:59:48,961][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 06:59:49,461][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 06:59:49,960][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 06:59:50,459][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 06:59:50,958][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 06:59:51,458][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 06:59:51,957][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 06:59:52,456][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 06:59:52,954][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 06:59:53,454][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 06:59:53,955][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 06:59:54,455][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 06:59:54,956][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 06:59:55,458][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 06:59:55,958][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 06:59:56,458][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 06:59:56,959][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 06:59:57,459][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 06:59:57,959][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 06:59:58,458][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 06:59:58,958][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 06:59:59,456][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 06:59:59,955][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:00:00,454][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:00:00,955][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:00:01,454][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:00:01,954][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:00:02,455][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:00:02,954][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:00:03,453][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:00:03,953][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:00:04,454][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:00:04,954][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:00:05,455][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:00:05,954][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:00:06,453][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:00:06,952][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:00:07,452][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:00:07,952][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:00:08,452][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:00:08,952][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:00:09,452][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:00:09,951][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:00:10,450][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:00:10,949][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:00:11,449][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:00:11,949][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:00:12,448][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:00:12,949][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:00:13,448][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 07:00:14,094][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 07:00:14,836][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:00:14,838][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:00:14,840][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:00:15,569][__main__][INFO] - Iteration 494 took 1m 14s (8.80% Gen, 90.21% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 24m 11s. Estimated total time: 61h 49m 45s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 39s, 500 more iterations: 10h 18m 17s. [2026-03-26 07:00:15,571][__main__][INFO] - Starting iteration 494. [2026-03-26 07:00:15,973][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:00:15,973][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:00:22,812][__main__][INFO] - Number of regex retries in iteration 494: 0 [2026-03-26 07:00:22,813][__main__][INFO] - agents played in iteration 494 are Alice, Bob [2026-03-26 07:00:23,692][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:00:23,737][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:00:23,738][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:00:23,739][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:00:24,338][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:00:24,794][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:00:25,301][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:00:25,801][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:00:26,300][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:00:26,800][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:00:27,300][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:00:27,799][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:00:28,299][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:00:28,799][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:00:29,298][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:00:29,798][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:00:30,297][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:00:30,797][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:00:31,296][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:00:31,796][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:00:32,295][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:00:32,794][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:00:33,295][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:00:33,795][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:00:34,295][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:00:34,794][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:00:35,293][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:00:35,792][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:00:36,290][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:00:36,791][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:00:37,291][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:00:37,791][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:00:38,291][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:00:38,790][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:00:39,289][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:00:39,788][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:00:40,288][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:00:40,787][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:00:41,285][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:00:41,785][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:00:42,284][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:00:42,782][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:00:43,281][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:00:43,780][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:00:44,279][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:00:44,778][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:00:45,276][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:00:45,775][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:00:46,274][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:00:46,773][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:00:47,271][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:00:47,769][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:00:48,271][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:00:48,773][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:00:49,273][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:00:49,774][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:00:50,274][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:00:50,775][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:00:51,275][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:00:51,775][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:00:52,276][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:00:52,775][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:00:53,273][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:00:53,771][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:00:54,269][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:00:54,767][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:00:55,266][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:00:55,764][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:00:56,263][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:00:56,763][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:00:57,262][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:00:57,762][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:00:58,262][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:00:58,761][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:00:59,261][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:00:59,761][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:01:00,262][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:01:00,762][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:01:01,263][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:01:01,763][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:01:02,263][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:01:02,762][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:01:03,263][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:01:03,763][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:01:04,263][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:01:04,762][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:01:05,262][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:01:05,761][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:01:06,260][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:01:06,760][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:01:07,260][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:01:07,760][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:01:08,260][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:01:08,760][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:01:09,259][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:01:09,759][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:01:10,258][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:01:10,757][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:01:11,257][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:01:11,756][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:01:12,255][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:01:12,753][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:01:13,252][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:01:13,753][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:01:14,253][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:01:14,753][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:01:15,252][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:01:15,754][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:01:16,255][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:01:16,753][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:01:17,253][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:01:17,752][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:01:18,254][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:01:18,753][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:01:19,251][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:01:19,749][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:01:20,247][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:01:20,746][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:01:21,245][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:01:21,745][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:01:22,246][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:01:22,745][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:01:23,245][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:01:23,744][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:01:24,244][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:01:24,744][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:01:25,244][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:01:25,743][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:01:26,241][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:01:26,740][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:01:27,238][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:01:27,737][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:01:28,236][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 07:01:28,862][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 07:01:29,654][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:01:29,657][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:01:29,658][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:01:30,364][__main__][INFO] - Iteration 495 took 1m 14s (9.19% Gen, 89.85% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 32m 48s. Estimated total time: 61h 59m 37s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 59s, 500 more iterations: 10h 19m 56s. [2026-03-26 07:01:30,366][__main__][INFO] - Starting iteration 495. [2026-03-26 07:01:30,767][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:01:30,767][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:01:38,291][__main__][INFO] - Number of regex retries in iteration 495: 0 [2026-03-26 07:01:38,292][__main__][INFO] - agents played in iteration 495 are Alice, Bob [2026-03-26 07:01:39,225][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:01:39,270][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:01:39,270][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:01:39,271][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:01:39,877][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:01:40,334][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:01:40,839][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:01:41,340][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:01:41,840][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:01:42,341][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:01:42,841][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:01:43,341][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:01:43,842][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:01:44,342][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:01:44,843][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:01:45,343][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:01:45,843][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:01:46,343][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:01:46,843][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:01:47,342][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:01:47,842][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:01:48,344][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:01:48,846][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:01:49,346][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:01:49,845][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:01:50,344][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:01:50,843][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:01:51,344][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:01:51,844][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:01:52,344][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:01:52,844][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:01:53,343][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:01:53,843][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:01:54,342][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:01:54,842][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:01:55,342][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:01:55,842][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:01:56,342][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:01:56,842][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:01:57,341][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:01:57,841][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:01:58,341][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:01:58,842][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:01:59,342][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:01:59,843][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:02:00,344][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:02:00,845][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:02:01,345][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:02:01,846][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:02:02,348][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:02:02,849][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:02:03,349][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:02:03,849][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:02:04,349][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:02:04,850][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:02:05,350][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:02:05,850][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:02:06,351][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:02:06,851][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:02:07,351][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:02:07,850][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:02:08,349][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:02:08,847][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:02:09,345][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:02:09,844][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:02:10,342][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:02:10,840][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:02:11,339][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:02:11,837][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:02:12,338][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:02:12,839][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:02:13,340][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:02:13,842][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:02:14,342][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:02:14,844][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:02:15,344][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:02:15,844][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:02:16,345][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:02:16,845][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:02:17,344][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:02:17,844][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:02:18,346][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:02:18,845][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:02:19,345][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:02:19,845][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:02:20,345][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:02:20,844][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:02:21,342][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:02:21,841][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:02:22,340][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:02:22,840][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:02:23,339][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:02:23,838][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:02:24,338][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:02:24,837][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:02:25,337][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:02:25,837][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:02:26,337][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:02:26,836][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:02:27,336][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:02:27,835][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:02:28,336][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:02:28,835][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:02:29,334][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:02:29,835][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:02:30,335][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:02:30,835][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:02:31,335][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:02:31,835][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:02:32,335][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:02:32,836][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:02:33,336][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:02:33,837][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:02:34,337][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:02:34,836][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:02:35,337][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:02:35,837][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:02:36,338][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:02:36,838][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:02:37,338][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:02:37,838][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:02:38,338][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:02:38,839][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:02:39,339][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:02:39,839][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:02:40,338][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:02:40,837][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:02:41,335][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:02:41,834][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:02:42,333][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:02:42,830][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:02:43,328][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:02:43,825][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21746 tokens. [2026-03-26 07:02:44,458][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:02:45,196][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:02:45,198][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:02:45,200][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:02:45,853][__main__][INFO] - Iteration 496 took 1m 15s (10.02% Gen, 89.11% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 6m 17s. Estimated total time: 62h 34m 21s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 8s, 500 more iterations: 10h 25m 43s. [2026-03-26 07:02:45,855][__main__][INFO] - Starting iteration 496. [2026-03-26 07:02:46,254][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:02:46,254][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:02:52,696][__main__][INFO] - Number of regex retries in iteration 496: 0 [2026-03-26 07:02:52,697][__main__][INFO] - agents played in iteration 496 are Alice, Bob [2026-03-26 07:02:53,634][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:02:53,679][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:02:53,679][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:02:53,680][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:02:54,284][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:02:54,739][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:02:55,245][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:02:55,744][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:02:56,245][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:02:56,745][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:02:57,246][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:02:57,746][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:02:58,246][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:02:58,744][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:02:59,243][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:02:59,742][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:03:00,242][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:03:00,739][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:03:01,238][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:03:01,737][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:03:02,237][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:03:02,737][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:03:03,237][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:03:03,737][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:03:04,236][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:03:04,736][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:03:05,236][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:03:05,737][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:03:06,237][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:03:06,738][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:03:07,238][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:03:07,737][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:03:08,237][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:03:08,736][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:03:09,236][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:03:09,736][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:03:10,236][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:03:10,736][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:03:11,237][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:03:11,737][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:03:12,238][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:03:12,738][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:03:13,239][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:03:13,739][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:03:14,239][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:03:14,739][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:03:15,238][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:03:15,738][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:03:16,238][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:03:16,738][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:03:17,238][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:03:17,738][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:03:18,240][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:03:18,739][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:03:19,239][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:03:19,738][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:03:20,237][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:03:20,736][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:03:21,236][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:03:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:03:22,235][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:03:22,733][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:03:23,232][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:03:23,729][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:03:24,227][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:03:24,726][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:03:25,225][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:03:25,723][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:03:26,221][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:03:26,721][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:03:27,220][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:03:27,720][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:03:28,222][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:03:28,722][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:03:29,222][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:03:29,721][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:03:30,221][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:03:30,720][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:03:31,218][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:03:31,716][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:03:32,217][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:03:32,716][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:03:33,215][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:03:33,714][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:03:34,212][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:03:34,711][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:03:35,210][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:03:35,708][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:03:36,208][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:03:36,707][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:03:37,207][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:03:37,706][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:03:38,205][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:03:38,705][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:03:39,204][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:03:39,704][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:03:40,204][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:03:40,703][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:03:41,203][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:03:41,702][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:03:42,202][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:03:42,702][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:03:43,201][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:03:43,701][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:03:44,201][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:03:44,700][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:03:45,199][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:03:45,698][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:03:46,198][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:03:46,695][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:03:47,194][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:03:47,693][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:03:48,194][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:03:48,692][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:03:49,190][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:03:49,688][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:03:50,186][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:03:50,683][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:03:51,182][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:03:51,681][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:03:52,180][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:03:52,678][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:03:53,177][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:03:53,676][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:03:54,174][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:03:54,672][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:03:55,170][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:03:55,669][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:03:56,168][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:03:56,667][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:03:57,165][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:03:57,664][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:03:58,163][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:03:58,789][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:03:59,529][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:03:59,532][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:03:59,533][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:04:00,181][__main__][INFO] - Iteration 497 took 1m 13s (8.71% Gen, 90.41% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 7m 4s. Estimated total time: 61h 36m 22s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 12s, 500 more iterations: 10h 16m 3s. [2026-03-26 07:04:00,183][__main__][INFO] - Starting iteration 497. [2026-03-26 07:04:00,584][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:04:00,585][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:04:07,642][__main__][INFO] - Number of regex retries in iteration 497: 0 [2026-03-26 07:04:07,643][__main__][INFO] - agents played in iteration 497 are Alice, Bob [2026-03-26 07:04:08,598][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:04:08,643][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:04:08,644][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:04:08,644][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:04:09,514][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:04:09,968][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:04:10,472][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:04:10,971][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:04:11,469][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:04:11,968][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:04:12,467][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:04:12,966][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:04:13,465][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:04:13,963][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:04:14,462][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:04:14,960][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:04:15,463][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:04:15,964][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:04:16,464][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:04:16,960][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:04:17,460][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:04:17,958][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:04:18,457][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:04:18,954][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:04:19,450][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:04:19,947][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:04:20,445][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:04:20,942][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:04:21,441][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:04:21,943][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:04:22,443][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:04:22,942][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:04:23,442][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:04:23,942][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:04:24,442][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:04:24,941][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:04:25,441][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:04:25,940][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:04:26,439][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:04:26,938][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:04:27,437][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:04:27,937][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:04:28,437][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:04:28,936][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:04:29,435][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:04:29,934][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:04:30,433][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:04:30,932][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:04:31,432][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:04:31,931][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:04:32,430][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:04:32,931][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:04:33,430][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:04:33,928][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:04:34,427][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:04:34,925][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:04:35,424][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:04:35,923][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:04:36,422][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:04:36,921][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:04:37,420][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:04:37,919][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:04:38,418][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:04:38,917][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:04:39,417][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:04:39,916][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:04:40,416][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:04:40,914][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:04:41,412][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:04:41,910][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:04:42,408][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:04:42,906][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:04:43,404][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:04:43,903][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:04:44,402][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:04:44,900][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:04:45,399][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:04:45,897][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:04:46,395][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:04:46,894][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:04:47,392][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:04:47,890][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:04:48,390][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:04:48,889][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:04:49,387][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:04:49,882][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:04:50,379][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:04:50,875][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:04:51,372][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:04:51,869][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:04:52,365][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:04:52,862][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:04:53,359][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:04:53,859][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:04:54,358][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:04:54,857][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:04:55,356][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:04:55,856][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:04:56,356][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:04:56,856][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:04:57,356][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:04:57,855][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:04:58,354][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:04:58,852][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:04:59,350][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:04:59,850][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:05:00,348][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:05:00,847][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:05:01,345][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:05:01,845][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:05:02,344][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:05:02,843][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:05:03,342][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:05:03,842][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:05:04,342][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:05:04,842][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:05:05,342][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:05:05,842][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:05:06,341][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:05:06,843][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:05:07,343][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:05:07,843][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:05:08,342][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:05:08,842][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:05:09,341][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:05:09,839][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:05:10,338][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:05:10,837][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:05:11,337][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:05:11,836][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:05:12,335][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:05:12,834][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:05:13,332][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21728 tokens. [2026-03-26 07:05:13,958][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.24%, ΔTime: 00:01:04 [2026-03-26 07:05:14,687][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:05:14,690][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:05:14,691][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:05:15,340][__main__][INFO] - Iteration 498 took 1m 14s (9.44% Gen, 89.69% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 51h 47m 17s. Estimated total time: 62h 17m 50s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 35s, 500 more iterations: 10h 22m 58s. [2026-03-26 07:05:15,342][__main__][INFO] - Starting iteration 498. [2026-03-26 07:05:15,746][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:05:15,747][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:05:22,552][__main__][INFO] - Number of regex retries in iteration 498: 0 [2026-03-26 07:05:22,553][__main__][INFO] - agents played in iteration 498 are Alice, Bob [2026-03-26 07:05:23,436][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:05:23,481][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:05:23,482][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:05:23,482][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:05:24,093][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:05:24,550][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:05:25,057][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:05:25,558][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:05:26,058][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:05:26,558][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:05:27,057][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:05:27,556][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:05:28,056][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:05:28,557][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:05:29,056][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:05:29,556][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:05:30,056][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:05:30,555][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:05:31,054][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:05:31,552][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:05:32,051][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:05:32,552][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:05:33,051][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:05:33,550][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:05:34,050][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:05:34,549][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:05:35,048][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:05:35,547][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:05:36,047][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:05:36,546][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:05:37,045][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:05:37,544][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:05:38,044][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:05:38,542][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:05:39,042][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:05:39,542][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:05:40,042][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:05:40,543][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:05:41,043][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:05:41,543][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:05:42,042][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:05:42,542][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:05:43,043][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:05:43,543][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:05:44,043][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:05:44,545][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:05:45,046][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:05:45,548][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:05:46,050][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:05:46,550][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:05:47,051][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:05:47,550][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:05:48,051][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:05:48,551][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:05:49,049][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:05:49,547][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:05:50,046][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:05:50,545][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:05:51,044][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:05:51,543][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:05:52,041][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:05:52,541][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:05:53,039][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:05:53,538][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:05:54,038][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:05:54,537][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:05:55,037][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:05:55,538][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:05:56,037][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:05:56,538][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:05:57,037][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:05:57,538][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:05:58,038][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:05:58,537][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:05:59,037][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:05:59,536][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:06:00,037][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:06:00,536][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:06:01,037][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:06:01,538][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:06:02,037][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:06:02,537][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:06:03,035][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:06:03,533][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:06:04,032][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:06:04,530][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:06:05,028][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:06:05,527][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:06:06,026][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:06:06,526][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:06:07,025][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:06:07,524][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:06:08,024][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:06:08,523][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:06:09,022][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:06:09,520][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:06:10,019][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:06:10,516][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:06:11,014][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:06:11,512][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:06:12,010][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:06:12,508][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:06:13,007][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:06:13,506][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:06:14,006][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:06:14,505][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:06:15,003][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:06:15,502][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:06:16,001][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:06:16,501][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:06:17,003][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:06:17,504][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:06:18,005][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:06:18,507][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:06:19,007][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:06:19,506][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:06:20,006][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:06:20,505][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:06:21,003][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:06:21,501][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:06:22,000][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:06:22,499][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:06:22,996][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:06:23,494][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:06:23,994][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:06:24,493][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:06:24,992][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:06:25,491][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:06:25,990][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:06:26,490][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:06:26,988][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:06:27,487][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:06:27,986][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 07:06:28,618][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 07:06:29,354][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:06:29,356][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:06:29,358][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:06:30,017][__main__][INFO] - Iteration 499 took 1m 14s (9.16% Gen, 89.95% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 21m 45s. Estimated total time: 61h 53m 33s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 47s, 500 more iterations: 10h 18m 55s. [2026-03-26 07:06:30,019][__main__][INFO] - Starting iteration 499. [2026-03-26 07:06:30,419][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:06:30,420][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:06:37,561][__main__][INFO] - Number of regex retries in iteration 499: 0 [2026-03-26 07:06:37,562][__main__][INFO] - agents played in iteration 499 are Alice, Bob [2026-03-26 07:06:38,496][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:06:38,541][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:06:38,542][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:06:38,542][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:06:39,150][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:06:39,604][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:06:40,107][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:06:40,606][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:06:41,104][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:06:41,603][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:06:42,101][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:06:42,600][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:06:43,100][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:06:43,602][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:06:44,102][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:06:44,602][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:06:45,102][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:06:45,601][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:06:46,100][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:06:46,601][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:06:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:06:47,600][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:06:48,100][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:06:48,600][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:06:49,099][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:06:49,598][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:06:50,096][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:06:50,597][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:06:51,096][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:06:51,597][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:06:52,096][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:06:52,594][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:06:53,094][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:06:53,592][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:06:54,092][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:06:54,591][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:06:55,091][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:06:55,591][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:06:56,090][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:06:56,589][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:06:57,088][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:06:57,586][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:06:58,085][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:06:58,585][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:06:59,084][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:06:59,583][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:07:00,082][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:07:00,581][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:07:01,080][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:07:01,580][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:07:02,080][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:07:02,580][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:07:03,080][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:07:03,580][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:07:04,080][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:07:04,580][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:07:05,081][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:07:05,581][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:07:06,082][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:07:06,582][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:07:07,083][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:07:07,582][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:07:08,082][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:07:08,581][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:07:09,081][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:07:09,580][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:07:10,079][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:07:10,579][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:07:11,077][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:07:11,577][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:07:12,076][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:07:12,575][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:07:13,076][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:07:13,575][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:07:14,073][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:07:14,571][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:07:15,070][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:07:15,571][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:07:16,073][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:07:16,573][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:07:17,072][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:07:17,572][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:07:18,071][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:07:18,572][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:07:19,071][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:07:19,570][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:07:20,069][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:07:20,567][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:07:21,068][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:07:21,567][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:07:22,065][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:07:22,564][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:07:23,063][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:07:23,563][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:07:24,063][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:07:24,562][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:07:25,062][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:07:25,562][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:07:26,060][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:07:26,559][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:07:27,058][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:07:27,558][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:07:28,057][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:07:28,557][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:07:29,057][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:07:29,556][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:07:30,054][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:07:30,553][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:07:31,051][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:07:31,552][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:07:32,051][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:07:32,550][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:07:33,049][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:07:33,549][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:07:34,048][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:07:34,546][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:07:35,046][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:07:35,547][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:07:36,047][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:07:36,547][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:07:37,046][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:07:37,546][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:07:38,046][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:07:38,546][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:07:39,047][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:07:39,547][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:07:40,046][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:07:40,545][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:07:41,045][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:07:41,544][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:07:42,044][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:07:42,544][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:07:43,044][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21725 tokens. [2026-03-26 07:07:43,686][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 07:07:44,427][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:07:44,430][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:07:44,431][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:07:45,084][__main__][INFO] - Iteration 500 took 1m 14s (9.57% Gen, 89.56% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 40m 13s. Estimated total time: 62h 13m 16s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 26s, 500 more iterations: 10h 22m 12s. [2026-03-26 07:07:45,086][__main__][INFO] - Starting iteration 500. [2026-03-26 07:07:45,489][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 9 and human policies 1. [2026-03-26 07:07:45,490][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:07:52,129][__main__][INFO] - Number of regex retries in iteration 500: 0 [2026-03-26 07:07:52,129][__main__][INFO] - agents played in iteration 500 are Alice, Bob [2026-03-26 07:07:53,068][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:07:53,113][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:07:53,114][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:07:53,114][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:07:53,731][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:07:54,187][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:07:54,691][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:07:55,192][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:07:55,692][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:07:56,192][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:07:56,691][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:07:57,191][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:07:57,692][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:07:58,187][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:07:58,683][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:07:59,180][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:07:59,677][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:08:00,175][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:08:00,671][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:08:01,167][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:08:01,663][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:08:02,164][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:08:02,664][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:08:03,163][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:08:03,663][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:08:04,162][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:08:04,662][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:08:05,161][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:08:05,660][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:08:06,158][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:08:06,657][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:08:07,156][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:08:07,655][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:08:08,155][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:08:08,656][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:08:09,156][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:08:09,656][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:08:10,160][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:08:10,664][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:08:11,166][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:08:11,668][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:08:12,170][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:08:12,673][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:08:13,174][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:08:13,676][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:08:14,178][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:08:14,678][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:08:15,179][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:08:15,678][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:08:16,177][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:08:16,677][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:08:17,177][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:08:17,677][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:08:18,182][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:08:18,684][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:08:19,185][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:08:19,685][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:08:20,185][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:08:20,686][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:08:21,187][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:08:21,688][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:08:22,187][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:08:22,688][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:08:23,188][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:08:23,690][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:08:24,190][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:08:24,691][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:08:25,193][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:08:25,693][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:08:26,193][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:08:26,693][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:08:27,192][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:08:27,692][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:08:28,192][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:08:28,692][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:08:29,191][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:08:29,691][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:08:30,187][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:08:30,683][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:08:31,179][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:08:31,676][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:08:32,173][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:08:32,669][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:08:33,165][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:08:33,662][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:08:34,164][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:08:34,663][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:08:35,163][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:08:35,663][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:08:36,163][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:08:36,663][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:08:37,163][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:08:37,663][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:08:38,162][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:08:38,662][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:08:39,161][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:08:39,661][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:08:40,161][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:08:40,661][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:08:41,160][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:08:41,659][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:08:42,163][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:08:42,666][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:08:43,168][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:08:43,670][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:08:44,173][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:08:44,675][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:08:45,178][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:08:45,685][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:08:46,187][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:08:46,688][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:08:47,187][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:08:47,687][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:08:48,189][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:08:48,689][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:08:49,189][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:08:49,690][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:08:50,190][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:08:50,690][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:08:51,190][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:08:51,690][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:08:52,189][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:08:52,689][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:08:53,189][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:08:53,689][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:08:54,189][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:08:54,690][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:08:55,190][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:08:55,691][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:08:56,191][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:08:56,692][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:08:57,192][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:08:57,692][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:08:58,335][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 07:08:59,076][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:08:59,079][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:08:59,080][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:09:02,947][__main__][INFO] - Iteration 501 took 1m 17s (8.57% Gen, 86.43% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 53h 58m 35s. Estimated total time: 64h 32m 56s. Time estimates for 10 more iterations: 12m 54s, 100 more iterations: 2h 9m 5s, 500 more iterations: 10h 45m 29s. [2026-03-26 07:09:02,950][__main__][INFO] - Starting iteration 501. [2026-03-26 07:09:03,349][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:09:03,350][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:09:03,956][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:09:10,178][__main__][INFO] - Number of regex retries in iteration 501: 1 [2026-03-26 07:09:10,179][__main__][INFO] - agents played in iteration 501 are Alice, Bob [2026-03-26 07:09:11,084][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:09:11,129][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:09:11,130][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:09:11,130][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:09:11,735][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:09:12,190][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:09:12,695][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:09:13,196][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:09:13,696][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:09:14,196][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:09:14,697][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:09:15,197][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:09:15,695][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:09:16,194][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:09:16,692][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:09:17,189][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:09:17,687][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:09:18,187][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:09:18,686][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:09:19,184][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:09:19,682][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:09:20,181][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:09:20,681][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:09:21,180][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:09:21,680][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:09:22,178][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:09:22,677][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:09:23,178][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:09:23,677][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:09:24,179][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:09:24,680][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:09:25,180][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:09:25,678][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:09:26,177][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:09:26,675][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:09:27,176][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:09:27,678][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:09:28,178][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:09:28,678][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:09:29,177][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:09:29,676][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:09:30,175][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:09:30,673][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:09:31,171][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:09:31,673][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:09:32,174][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:09:32,674][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:09:33,174][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:09:33,674][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:09:34,172][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:09:34,673][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:09:35,173][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:09:35,673][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:09:36,172][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:09:36,670][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:09:37,169][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:09:37,668][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:09:38,166][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:09:38,664][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:09:39,164][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:09:39,663][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:09:40,162][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:09:40,661][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:09:41,160][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:09:41,659][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:09:42,157][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:09:42,658][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:09:43,158][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:09:43,659][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:09:44,158][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:09:44,657][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:09:45,155][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:09:45,653][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:09:46,152][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:09:46,650][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:09:47,149][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:09:47,649][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:09:48,149][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:09:48,648][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:09:49,146][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:09:49,644][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:09:50,142][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:09:50,641][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:09:51,139][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:09:51,638][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:09:52,137][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:09:52,635][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:09:53,133][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:09:53,631][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:09:54,129][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:09:54,627][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:09:55,127][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:09:55,627][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:09:56,126][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:09:56,624][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:09:57,123][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:09:57,622][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:09:58,121][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:09:58,620][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:09:59,120][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:09:59,619][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:10:00,118][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:10:00,618][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:10:01,118][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:10:01,618][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:10:02,118][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:10:02,618][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:10:03,118][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:10:03,620][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:10:04,121][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:10:04,622][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:10:05,123][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:10:05,625][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:10:06,126][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:10:06,626][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:10:07,127][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:10:07,629][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:10:08,128][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:10:08,628][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:10:09,127][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:10:09,628][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:10:10,128][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:10:10,628][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:10:11,126][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:10:11,629][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:10:12,129][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:10:12,633][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:10:13,134][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:10:13,634][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:10:14,135][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:10:14,635][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:10:15,136][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:10:15,645][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:10:16,331][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 07:10:17,065][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:10:17,067][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:10:17,069][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:10:17,870][__main__][INFO] - Iteration 502 took 1m 14s (9.16% Gen, 89.76% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 30m 29s. Estimated total time: 62h 6m 5s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 12s, 500 more iterations: 10h 21m 0s. [2026-03-26 07:10:17,872][__main__][INFO] - Starting iteration 502. [2026-03-26 07:10:18,275][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:10:18,275][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:10:25,152][__main__][INFO] - Number of regex retries in iteration 502: 0 [2026-03-26 07:10:25,153][__main__][INFO] - agents played in iteration 502 are Alice, Bob [2026-03-26 07:10:26,079][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:10:26,124][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:10:26,125][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:10:26,126][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:10:26,732][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:10:27,188][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:10:27,691][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:10:28,191][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:10:28,689][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:10:29,189][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:10:29,689][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:10:30,189][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:10:30,688][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:10:31,188][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:10:31,688][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:10:32,187][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:10:32,687][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:10:33,188][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:10:33,688][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:10:34,187][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:10:34,687][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:10:35,188][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:10:35,687][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:10:36,187][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:10:36,686][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:10:37,186][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:10:37,686][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:10:38,185][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:10:38,685][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:10:39,184][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:10:39,684][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:10:40,183][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:10:40,684][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:10:41,184][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:10:41,684][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:10:42,184][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:10:42,683][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:10:43,180][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:10:43,677][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:10:44,173][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:10:44,671][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:10:45,169][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:10:45,667][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:10:46,165][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:10:46,663][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:10:47,163][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:10:47,662][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:10:48,164][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:10:48,665][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:10:49,165][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:10:49,664][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:10:50,163][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:10:50,663][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:10:51,162][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:10:51,662][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:10:52,161][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:10:52,661][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:10:53,160][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:10:53,659][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:10:54,158][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:10:54,657][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:10:55,158][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:10:55,658][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:10:56,160][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:10:56,661][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:10:57,161][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:10:57,662][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:10:58,162][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:10:58,663][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:10:59,163][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:10:59,662][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:11:00,162][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:11:00,662][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:11:01,161][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:11:01,660][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:11:02,159][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:11:02,659][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:11:03,159][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:11:03,659][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:11:04,159][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:11:04,659][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:11:05,158][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:11:05,657][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:11:06,156][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:11:06,659][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:11:07,159][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:11:07,660][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:11:08,160][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:11:08,660][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:11:09,161][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:11:09,662][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:11:10,162][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:11:10,662][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:11:11,162][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:11:11,663][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:11:12,163][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:11:12,665][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:11:13,166][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:11:13,667][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:11:14,167][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:11:14,667][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:11:15,164][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:11:15,661][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:11:16,157][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:11:16,654][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:11:17,151][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:11:17,649][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:11:18,148][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:11:18,648][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:11:19,148][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:11:19,647][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:11:20,146][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:11:20,645][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:11:21,145][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:11:21,644][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:11:22,144][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:11:22,644][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:11:23,143][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:11:23,643][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:11:24,143][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:11:24,642][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:11:25,143][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:11:25,642][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:11:26,142][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:11:26,641][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:11:27,142][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:11:27,642][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:11:28,143][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:11:28,644][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:11:29,145][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:11:29,645][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:11:30,146][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:11:30,646][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:11:31,301][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 07:11:32,035][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:11:32,037][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:11:32,039][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:11:32,688][__main__][INFO] - Iteration 503 took 1m 14s (9.24% Gen, 89.88% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 23m 51s. Estimated total time: 62h 0m 42s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 1s, 500 more iterations: 10h 20m 7s. [2026-03-26 07:11:32,690][__main__][INFO] - Starting iteration 503. [2026-03-26 07:11:33,091][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:11:33,091][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:11:34,282][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:11:40,707][__main__][INFO] - Number of regex retries in iteration 503: 1 [2026-03-26 07:11:40,708][__main__][INFO] - agents played in iteration 503 are Alice, Bob [2026-03-26 07:11:41,641][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:11:41,685][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:11:41,686][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:11:41,687][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:11:42,285][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:11:42,741][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:11:43,243][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:11:43,742][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:11:44,241][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:11:44,739][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:11:45,239][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:11:45,741][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:11:46,240][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:11:46,746][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:11:47,246][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:11:47,746][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:11:48,249][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:11:48,750][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:11:49,250][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:11:49,750][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:11:50,250][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:11:50,749][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:11:51,248][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:11:51,748][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:11:52,248][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:11:52,747][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:11:53,245][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:11:53,745][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:11:54,244][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:11:54,744][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:11:55,245][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:11:55,745][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:11:56,245][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:11:56,744][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:11:57,243][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:11:57,743][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:11:58,243][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:11:58,743][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:11:59,243][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:11:59,743][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:12:00,243][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:12:00,743][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:12:01,243][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:12:01,743][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:12:02,243][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:12:02,742][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:12:03,242][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:12:03,742][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:12:04,242][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:12:04,742][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:12:05,241][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:12:05,742][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:12:06,243][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:12:06,742][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:12:07,242][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:12:07,741][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:12:08,241][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:12:08,741][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:12:09,241][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:12:09,740][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:12:10,240][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:12:10,739][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:12:11,239][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:12:11,739][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:12:12,239][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:12:12,741][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:12:13,240][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:12:13,742][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:12:14,241][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:12:14,741][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:12:15,239][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:12:15,737][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:12:16,238][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:12:16,738][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:12:17,236][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:12:17,735][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:12:18,235][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:12:18,735][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:12:19,235][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:12:19,734][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:12:20,235][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:12:20,736][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:12:21,236][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:12:21,735][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:12:22,235][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:12:22,740][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:12:23,242][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:12:23,742][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:12:24,243][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:12:24,743][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:12:25,244][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:12:25,744][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:12:26,245][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:12:26,752][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:12:27,254][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:12:27,754][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:12:28,255][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:12:28,755][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:12:29,256][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:12:29,759][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:12:30,261][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:12:30,765][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:12:31,267][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:12:31,768][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:12:32,270][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:12:32,771][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:12:33,273][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:12:33,773][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:12:34,274][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:12:34,778][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:12:35,281][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:12:35,783][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:12:36,285][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:12:36,787][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:12:37,289][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:12:37,790][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:12:38,291][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:12:38,797][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:12:39,298][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:12:39,800][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:12:40,301][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:12:40,803][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:12:41,303][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:12:41,804][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:12:42,306][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:12:42,811][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:12:43,312][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:12:43,814][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:12:44,316][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:12:44,819][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:12:45,320][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:12:45,822][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:12:46,331][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 07:12:46,960][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-26 07:12:47,693][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:12:47,696][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:12:47,697][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:12:48,347][__main__][INFO] - Iteration 504 took 1m 15s (10.12% Gen, 89.01% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 52h 4m 45s. Estimated total time: 62h 42m 51s. Time estimates for 10 more iterations: 12m 32s, 100 more iterations: 2h 5m 25s, 500 more iterations: 10h 27m 8s. [2026-03-26 07:12:48,350][__main__][INFO] - Starting iteration 504. [2026-03-26 07:12:48,750][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:12:48,750][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:12:49,347][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:12:56,227][__main__][INFO] - Number of regex retries in iteration 504: 1 [2026-03-26 07:12:56,227][__main__][INFO] - agents played in iteration 504 are Alice, Bob [2026-03-26 07:12:57,171][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:12:57,217][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:12:57,218][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:12:57,218][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:12:57,822][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:12:58,279][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:12:58,784][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:12:59,285][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:12:59,786][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:13:00,288][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:13:00,788][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:13:01,288][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:13:01,789][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:13:02,288][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:13:02,788][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:13:03,288][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:13:03,788][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:13:04,288][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:13:04,787][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:13:05,287][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:13:05,787][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:13:06,289][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:13:06,789][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:13:07,288][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:13:07,789][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:13:08,288][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:13:08,788][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:13:09,288][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:13:09,788][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:13:10,287][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:13:10,788][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:13:11,287][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:13:11,790][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:13:12,289][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:13:12,788][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:13:13,288][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:13:13,787][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:13:14,289][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:13:14,790][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:13:15,291][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:13:15,794][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:13:16,296][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:13:16,796][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:13:17,296][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:13:17,797][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:13:18,299][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:13:18,799][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:13:19,298][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:13:19,799][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:13:20,299][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:13:20,800][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:13:21,300][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:13:21,800][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:13:22,299][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:13:22,799][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:13:23,301][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:13:23,800][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:13:24,300][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:13:24,800][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:13:25,302][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:13:25,802][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:13:26,303][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:13:26,803][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:13:27,303][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:13:27,803][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:13:28,304][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:13:28,804][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:13:29,304][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:13:29,805][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:13:30,305][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:13:30,804][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:13:31,305][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:13:31,805][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:13:32,305][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:13:32,805][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:13:33,305][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:13:33,806][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:13:34,306][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:13:34,806][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:13:35,305][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:13:35,805][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:13:36,304][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:13:36,802][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:13:37,303][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:13:37,802][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:13:38,303][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:13:38,808][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:13:39,308][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:13:39,809][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:13:40,310][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:13:40,811][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:13:41,312][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:13:41,813][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:13:42,313][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:13:42,814][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:13:43,315][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:13:43,816][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:13:44,317][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:13:44,818][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:13:45,318][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:13:45,818][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:13:46,320][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:13:46,824][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:13:47,326][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:13:47,830][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:13:48,334][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:13:48,836][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:13:49,341][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:13:49,844][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:13:50,347][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:13:50,850][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:13:51,351][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:13:51,854][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:13:52,356][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:13:52,857][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:13:53,360][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:13:53,862][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:13:54,365][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:13:54,868][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:13:55,372][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:13:55,874][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:13:56,376][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:13:56,877][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:13:57,379][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:13:57,883][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:13:58,387][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:13:58,889][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:13:59,392][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:13:59,895][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:14:00,398][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:14:00,901][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:14:01,404][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:14:01,911][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21740 tokens. [2026-03-26 07:14:02,572][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 07:14:03,321][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:14:03,323][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:14:03,325][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:14:04,144][__main__][INFO] - Iteration 505 took 1m 15s (9.92% Gen, 88.99% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 52h 10m 23s. Estimated total time: 62h 49m 45s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 39s, 500 more iterations: 10h 28m 17s. [2026-03-26 07:14:04,146][__main__][INFO] - Starting iteration 505. [2026-03-26 07:14:04,551][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:14:04,552][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:14:09,463][mllm.models.large_language_model_local][WARNING] - Response Proposals often benefit from a mix of aggression and cooperation, aiming to maximize your own points while accounting for the proportional distribution if things go over. Given the values, wearing the balls (10) is significantly more valuable for you than for Alice (1 for hats and books, 1 for balls), so you should propose to take most of the balls. Alice is more inclined to wear hats and books, so you should share those. Here’s a strategic distribution: - Hats: 5 (to balance Alice’s value and prevent a proportional distribution) - Books: 5 (same logic as hats) - Balls: 10 (maximum for you, as you value them the most) So, the proposal would be: ``` Proposal: 5 hats, 5 books, 10 balls ``` did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:14:12,524][__main__][INFO] - Number of regex retries in iteration 505: 1 [2026-03-26 07:14:12,524][__main__][INFO] - agents played in iteration 505 are Alice, Bob [2026-03-26 07:14:13,475][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:14:13,522][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:14:13,523][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:14:13,524][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:14:14,136][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:14:14,593][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:14:15,098][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:14:15,600][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:14:16,103][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:14:16,605][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:14:17,105][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:14:17,607][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:14:18,108][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:14:18,611][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:14:19,111][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:14:19,611][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:14:20,110][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:14:20,610][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:14:21,109][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:14:21,608][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:14:22,108][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:14:22,606][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:14:23,105][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:14:23,604][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:14:24,102][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:14:24,601][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:14:25,100][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:14:25,600][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:14:26,098][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:14:26,594][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:14:27,092][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:14:27,589][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:14:28,086][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:14:28,583][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:14:29,080][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:14:29,577][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:14:30,074][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:14:30,573][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:14:31,072][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:14:31,571][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:14:32,072][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:14:32,570][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:14:33,068][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:14:33,567][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:14:34,066][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:14:34,566][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:14:35,065][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:14:35,564][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:14:36,064][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:14:36,563][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:14:37,063][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:14:37,562][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:14:38,061][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:14:38,560][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:14:39,060][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:14:39,559][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:14:40,057][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:14:40,555][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:14:41,054][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:14:41,552][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:14:42,051][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:14:42,550][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:14:43,050][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:14:43,549][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:14:44,048][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:14:44,547][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:14:45,046][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:14:45,549][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:14:46,051][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:14:46,552][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:14:47,052][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:14:47,552][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:14:48,053][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:14:48,555][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:14:49,055][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:14:49,555][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:14:50,055][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:14:50,554][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:14:51,054][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:14:51,554][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:14:52,053][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:14:52,552][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:14:53,052][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:14:53,552][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:14:54,052][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:14:54,551][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:14:55,050][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:14:55,548][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:14:56,047][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:14:56,546][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:14:57,044][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:14:57,544][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:14:58,043][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:14:58,539][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:14:59,035][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:14:59,531][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:15:00,028][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:15:00,526][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:15:01,024][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:15:01,522][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:15:02,019][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:15:02,519][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:15:03,018][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:15:03,516][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:15:04,013][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:15:04,511][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:15:05,010][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:15:05,508][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:15:06,007][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:15:06,506][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:15:07,005][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:15:07,503][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:15:08,003][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:15:08,503][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:15:09,002][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:15:09,502][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:15:10,001][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:15:10,500][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:15:10,999][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:15:11,498][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:15:11,997][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:15:12,496][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:15:12,996][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:15:13,495][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:15:14,000][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:15:14,501][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:15:15,001][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:15:15,501][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:15:16,001][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:15:16,501][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:15:17,001][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:15:17,502][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:15:18,004][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 07:15:18,669][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 07:15:19,545][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:15:19,548][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:15:19,549][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:15:20,265][__main__][INFO] - Iteration 506 took 1m 15s (10.53% Gen, 88.52% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 52h 25m 4s. Estimated total time: 63h 5m 42s. Time estimates for 10 more iterations: 12m 37s, 100 more iterations: 2h 6m 11s, 500 more iterations: 10h 30m 57s. [2026-03-26 07:15:20,267][__main__][INFO] - Starting iteration 506. [2026-03-26 07:15:20,665][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:15:20,666][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:15:23,296][mllm.models.large_language_model_local][WARNING] - Response Proposal: x hats, 10 books, 10 balls Explanation: Given the per-item values, I value balls the most (10), followed by books (1), and hats have the least value (1) to Alice. Therefore, it is in my best interest to claim as many balls and books as possible, while leaving all the hats for Alice. This strategy maximizes my potential points from the highly valued items. did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:15:28,207][__main__][INFO] - Number of regex retries in iteration 506: 1 [2026-03-26 07:15:28,208][__main__][INFO] - agents played in iteration 506 are Alice, Bob [2026-03-26 07:15:29,092][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:15:29,137][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:15:29,138][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:15:29,138][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:15:29,750][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:15:30,205][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:15:30,708][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:15:31,207][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:15:31,706][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:15:32,208][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:15:32,707][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:15:33,207][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:15:33,707][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:15:34,210][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:15:34,713][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:15:35,215][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:15:35,717][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:15:36,222][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:15:36,725][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:15:37,228][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:15:37,731][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:15:38,232][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:15:38,731][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:15:39,232][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:15:39,732][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:15:40,235][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:15:40,736][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:15:41,237][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:15:41,739][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:15:42,243][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:15:42,746][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:15:43,249][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:15:43,762][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:15:44,266][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:15:44,768][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:15:45,271][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:15:45,774][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:15:46,276][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:15:46,777][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:15:47,276][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:15:47,776][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:15:48,279][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:15:48,781][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:15:49,281][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:15:49,781][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:15:50,283][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:15:50,786][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:15:51,288][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:15:51,792][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:15:52,293][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:15:52,795][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:15:53,296][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:15:53,797][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:15:54,298][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:15:54,799][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:15:55,301][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:15:55,801][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:15:56,302][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:15:56,803][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:15:57,303][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:15:57,804][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:15:58,304][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:15:58,804][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:15:59,304][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:15:59,804][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:16:00,304][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:16:00,805][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:16:01,304][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:16:01,804][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:16:02,304][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:16:02,803][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:16:03,304][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:16:03,803][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:16:04,304][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:16:04,804][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:16:05,303][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:16:05,802][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:16:06,304][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:16:06,805][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:16:07,306][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:16:07,807][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:16:08,308][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:16:08,809][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:16:09,310][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:16:09,811][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:16:10,309][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:16:10,808][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:16:11,308][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:16:11,807][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:16:12,306][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:16:12,806][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:16:13,305][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:16:13,804][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:16:14,304][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:16:14,806][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:16:15,309][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:16:15,812][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:16:16,314][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:16:16,814][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:16:17,315][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:16:17,816][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:16:18,317][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:16:18,817][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:16:19,315][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:16:19,814][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:16:20,312][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:16:20,810][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:16:21,309][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:16:21,809][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:16:22,308][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:16:22,808][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:16:23,307][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:16:23,807][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:16:24,306][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:16:24,806][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:16:25,309][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:16:25,809][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:16:26,309][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:16:26,809][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:16:27,308][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:16:27,807][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:16:28,307][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:16:28,807][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:16:29,307][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:16:29,806][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:16:30,306][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:16:30,806][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:16:31,304][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:16:31,804][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:16:32,303][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:16:32,803][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:16:33,302][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:16:33,804][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21680 tokens. [2026-03-26 07:16:34,472][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 07:16:35,935][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:16:35,938][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:16:35,939][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:16:36,646][__main__][INFO] - Iteration 507 took 1m 15s (9.93% Gen, 89.14% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 52h 37m 8s. Estimated total time: 63h 19m 3s. Time estimates for 10 more iterations: 12m 39s, 100 more iterations: 2h 6m 38s, 500 more iterations: 10h 33m 10s. [2026-03-26 07:16:36,648][__main__][INFO] - Starting iteration 507. [2026-03-26 07:16:37,051][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:16:37,052][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:16:37,648][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:16:41,185][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:16:43,915][__main__][INFO] - Number of regex retries in iteration 507: 2 [2026-03-26 07:16:43,916][__main__][INFO] - agents played in iteration 507 are Alice, Bob [2026-03-26 07:16:44,846][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:16:44,892][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:16:44,893][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:16:44,893][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:16:45,506][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:16:45,963][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:16:46,468][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:16:46,970][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:16:47,470][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:16:47,974][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:16:48,479][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:16:48,980][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:16:49,482][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:16:49,985][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:16:50,488][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:16:50,991][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:16:51,492][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:16:51,997][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:16:52,499][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:16:53,001][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:16:53,504][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:16:54,005][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:16:54,505][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:16:55,006][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:16:55,507][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:16:56,010][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:16:56,511][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:16:57,012][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:16:57,513][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:16:58,015][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:16:58,516][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:16:59,017][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:16:59,518][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:17:00,023][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:17:00,525][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:17:01,028][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:17:01,529][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:17:02,031][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:17:02,534][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:17:03,037][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:17:03,545][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:17:04,049][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:17:04,552][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:17:05,055][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:17:05,558][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:17:06,060][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:17:06,562][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:17:07,063][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:17:07,577][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:17:08,079][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:17:08,581][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:17:09,083][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:17:09,583][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:17:10,085][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:17:10,586][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:17:11,087][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:17:11,591][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:17:12,092][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:17:12,592][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:17:13,093][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:17:13,593][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:17:14,094][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:17:14,595][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:17:15,096][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:17:15,597][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:17:16,098][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:17:16,599][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:17:17,102][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:17:17,603][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:17:18,105][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:17:18,608][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:17:19,109][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:17:19,611][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:17:20,111][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:17:20,612][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:17:21,112][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:17:21,613][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:17:22,113][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:17:22,613][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:17:23,113][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:17:23,614][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:17:24,114][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:17:24,615][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:17:25,116][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:17:25,617][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:17:26,116][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:17:26,614][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:17:27,113][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:17:27,612][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:17:28,112][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:17:28,610][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:17:29,108][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:17:29,607][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:17:30,107][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:17:30,607][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:17:31,108][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:17:31,608][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:17:32,108][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:17:32,607][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:17:33,107][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:17:33,606][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:17:34,107][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:17:34,608][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:17:35,109][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:17:35,610][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:17:36,110][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:17:36,610][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:17:37,109][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:17:37,610][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:17:38,110][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:17:38,610][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:17:39,109][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:17:39,608][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:17:40,107][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:17:40,607][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:17:41,106][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:17:41,605][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:17:42,104][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:17:42,604][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:17:43,102][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:17:43,601][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:17:44,099][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:17:44,597][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:17:45,095][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:17:45,597][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:17:46,100][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:17:46,601][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:17:47,101][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:17:47,601][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:17:48,102][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:17:48,603][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:17:49,103][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:17:49,603][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:17:50,254][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 07:17:50,993][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:17:50,995][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:17:50,997][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:17:51,711][__main__][INFO] - Iteration 508 took 1m 14s (9.19% Gen, 89.85% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 51h 29m 51s. Estimated total time: 62h 13m 1s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 26s, 500 more iterations: 10h 22m 10s. [2026-03-26 07:17:51,713][__main__][INFO] - Starting iteration 508. [2026-03-26 07:17:52,113][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:17:52,113][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:17:59,047][__main__][INFO] - Number of regex retries in iteration 508: 0 [2026-03-26 07:17:59,048][__main__][INFO] - agents played in iteration 508 are Alice, Bob [2026-03-26 07:17:59,980][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:18:00,025][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:18:00,026][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:18:00,026][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:18:00,635][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:18:01,093][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:18:01,598][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:18:02,098][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:18:02,598][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:18:03,100][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:18:03,600][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:18:04,101][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:18:04,601][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:18:05,101][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:18:05,601][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:18:06,101][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:18:06,602][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:18:07,102][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:18:07,603][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:18:08,106][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:18:08,607][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:18:09,107][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:18:09,606][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:18:10,106][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:18:10,605][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:18:11,104][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:18:11,604][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:18:12,103][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:18:12,602][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:18:13,101][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:18:13,601][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:18:14,099][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:18:14,597][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:18:15,096][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:18:15,595][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:18:16,093][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:18:16,591][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:18:17,089][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:18:17,587][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:18:18,086][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:18:18,587][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:18:19,086][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:18:19,583][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:18:20,083][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:18:20,581][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:18:21,082][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:18:21,583][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:18:22,082][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:18:22,581][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:18:23,081][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:18:23,582][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:18:24,083][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:18:24,582][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:18:25,083][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:18:25,583][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:18:26,083][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:18:26,582][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:18:27,081][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:18:27,579][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:18:28,080][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:18:28,581][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:18:29,079][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:18:29,577][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:18:30,075][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:18:30,573][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:18:31,070][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:18:31,568][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:18:32,067][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:18:32,564][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:18:33,065][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:18:33,565][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:18:34,065][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:18:34,565][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:18:35,066][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:18:35,566][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:18:36,067][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:18:36,567][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:18:37,068][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:18:37,568][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:18:38,069][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:18:38,570][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:18:39,070][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:18:39,571][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:18:40,071][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:18:40,571][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:18:41,071][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:18:41,570][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:18:42,069][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:18:42,569][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:18:43,068][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:18:43,567][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:18:44,067][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:18:44,567][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:18:45,065][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:18:45,564][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:18:46,064][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:18:46,563][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:18:47,062][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:18:47,561][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:18:48,061][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:18:48,563][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:18:49,061][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:18:49,560][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:18:50,058][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:18:50,557][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:18:51,055][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:18:51,554][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:18:52,053][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:18:52,551][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:18:53,050][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:18:53,550][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:18:54,051][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:18:54,551][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:18:55,052][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:18:55,552][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:18:56,052][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:18:56,552][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:18:57,052][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:18:57,551][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:18:58,052][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:18:58,552][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:18:59,051][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:18:59,551][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:19:00,050][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:19:00,549][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:19:01,048][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:19:01,547][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:19:02,044][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:19:02,542][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:19:03,040][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:19:03,538][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:19:04,035][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:19:04,532][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 07:19:05,166][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:19:05,898][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:19:05,900][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:19:05,902][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:19:06,552][__main__][INFO] - Iteration 509 took 1m 14s (9.32% Gen, 89.81% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 17m 37s. Estimated total time: 62h 2m 2s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 4s, 500 more iterations: 10h 20m 20s. [2026-03-26 07:19:06,554][__main__][INFO] - Starting iteration 509. [2026-03-26 07:19:06,953][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:19:06,954][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:19:14,345][__main__][INFO] - Number of regex retries in iteration 509: 0 [2026-03-26 07:19:14,346][__main__][INFO] - agents played in iteration 509 are Alice, Bob [2026-03-26 07:19:15,294][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:19:15,339][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:19:15,340][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:19:15,341][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:19:15,947][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:19:16,402][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:19:16,906][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:19:17,406][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:19:17,905][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:19:18,406][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:19:18,906][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:19:19,406][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:19:19,905][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:19:20,406][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:19:20,908][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:19:21,414][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:19:21,915][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:19:22,415][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:19:22,916][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:19:23,416][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:19:23,916][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:19:24,420][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:19:24,922][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:19:25,424][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:19:25,925][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:19:26,425][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:19:26,926][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:19:27,427][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:19:27,928][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:19:28,429][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:19:28,930][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:19:29,430][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:19:29,929][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:19:30,428][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:19:30,928][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:19:31,428][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:19:31,929][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:19:32,427][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:19:32,925][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:19:33,423][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:19:33,921][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:19:34,419][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:19:34,917][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:19:35,416][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:19:35,913][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:19:36,412][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:19:36,911][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:19:37,409][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:19:37,908][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:19:38,407][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:19:38,905][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:19:39,404][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:19:39,905][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:19:40,404][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:19:40,905][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:19:41,404][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:19:41,904][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:19:42,403][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:19:42,903][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:19:43,403][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:19:43,903][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:19:44,402][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:19:44,901][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:19:45,398][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:19:45,897][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:19:46,394][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:19:46,893][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:19:47,391][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:19:47,890][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:19:48,391][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:19:48,890][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:19:49,388][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:19:49,887][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:19:50,386][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:19:50,886][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:19:51,385][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:19:51,884][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:19:52,385][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:19:52,885][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:19:53,385][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:19:53,887][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:19:54,388][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:19:54,889][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:19:55,390][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:19:55,891][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:19:56,391][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:19:56,891][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:19:57,391][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:19:57,891][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:19:58,392][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:19:58,892][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:19:59,392][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:19:59,892][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:20:00,392][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:20:00,891][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:20:01,390][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:20:01,889][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:20:02,389][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:20:02,888][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:20:03,387][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:20:03,886][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:20:04,385][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:20:04,884][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:20:05,383][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:20:05,882][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:20:06,381][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:20:06,879][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:20:07,377][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:20:07,875][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:20:08,374][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:20:08,872][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:20:09,371][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:20:09,870][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:20:10,369][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:20:10,868][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:20:11,366][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:20:11,864][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:20:12,365][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:20:12,864][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:20:13,365][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:20:13,865][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:20:14,366][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:20:14,865][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:20:15,365][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:20:15,865][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:20:16,364][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:20:16,862][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:20:17,362][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:20:17,861][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:20:18,361][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:20:18,860][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:20:19,358][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:20:19,856][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 07:20:20,485][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:20:21,223][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:20:21,225][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:20:21,227][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:20:21,886][__main__][INFO] - Iteration 510 took 1m 14s (9.86% Gen, 89.25% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 40m 59s. Estimated total time: 62h 26m 39s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 53s, 500 more iterations: 10h 24m 26s. [2026-03-26 07:20:21,888][__main__][INFO] - Starting iteration 510. [2026-03-26 07:20:22,287][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:20:22,288][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:20:29,134][__main__][INFO] - Number of regex retries in iteration 510: 0 [2026-03-26 07:20:29,135][__main__][INFO] - agents played in iteration 510 are Alice, Bob [2026-03-26 07:20:30,010][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:20:30,055][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:20:30,056][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:20:30,056][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:20:30,660][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:20:31,116][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:20:31,622][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:20:32,122][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:20:32,623][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:20:33,123][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:20:33,623][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:20:34,123][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:20:34,622][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:20:35,122][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:20:35,621][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:20:36,122][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:20:36,621][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:20:37,120][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:20:37,618][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:20:38,115][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:20:38,613][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:20:39,113][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:20:39,614][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:20:40,113][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:20:40,613][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:20:41,113][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:20:41,613][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:20:42,112][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:20:42,611][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:20:43,110][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:20:43,609][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:20:44,108][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:20:44,606][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:20:45,104][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:20:45,607][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:20:46,108][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:20:46,607][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:20:47,109][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:20:47,610][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:20:48,111][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:20:48,612][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:20:49,111][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:20:49,610][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:20:50,110][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:20:50,611][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:20:51,111][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:20:51,611][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:20:52,111][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:20:52,611][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:20:53,111][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:20:53,611][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:20:54,111][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:20:54,611][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:20:55,111][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:20:55,611][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:20:56,110][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:20:56,609][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:20:57,109][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:20:57,608][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:20:58,107][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:20:58,607][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:20:59,107][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:20:59,607][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:21:00,106][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:21:00,607][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:21:01,107][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:21:01,607][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:21:02,107][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:21:02,608][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:21:03,107][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:21:03,606][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:21:04,106][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:21:04,605][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:21:05,105][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:21:05,604][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:21:06,102][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:21:06,602][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:21:07,100][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:21:07,598][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:21:08,097][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:21:08,596][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:21:09,094][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:21:09,592][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:21:10,090][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:21:10,588][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:21:11,088][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:21:11,587][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:21:12,088][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:21:12,589][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:21:13,090][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:21:13,589][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:21:14,089][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:21:14,588][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:21:15,088][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:21:15,587][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:21:16,086][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:21:16,585][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:21:17,084][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:21:17,583][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:21:18,082][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:21:18,584][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:21:19,085][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:21:19,585][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:21:20,086][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:21:20,587][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:21:21,087][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:21:21,586][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:21:22,087][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:21:22,588][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:21:23,088][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:21:23,589][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:21:24,089][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:21:24,589][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:21:25,089][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:21:25,589][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:21:26,089][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:21:26,589][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:21:27,088][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:21:27,587][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:21:28,086][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:21:28,584][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:21:29,083][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:21:29,581][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:21:30,080][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:21:30,580][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:21:31,080][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:21:31,580][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:21:32,080][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:21:32,580][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:21:33,080][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:21:33,581][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:21:34,082][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:21:34,584][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 07:21:35,228][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:21:35,976][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:21:35,979][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:21:35,980][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:21:36,832][__main__][INFO] - Iteration 511 took 1m 14s (9.19% Gen, 89.67% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 20m 23s. Estimated total time: 62h 7m 17s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 14s, 500 more iterations: 10h 21m 12s. [2026-03-26 07:21:36,834][__main__][INFO] - Starting iteration 511. [2026-03-26 07:21:37,236][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:21:37,237][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:21:43,848][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 0 books, 30 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:21:44,888][__main__][INFO] - Number of regex retries in iteration 511: 1 [2026-03-26 07:21:44,889][__main__][INFO] - agents played in iteration 511 are Alice, Bob [2026-03-26 07:21:45,826][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:21:45,871][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:21:45,871][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:21:45,872][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:21:46,472][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:21:46,926][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:21:47,430][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:21:47,930][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:21:48,433][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:21:48,933][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:21:49,432][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:21:49,931][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:21:50,430][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:21:50,929][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:21:51,427][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:21:51,926][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:21:52,424][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:21:52,923][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:21:53,423][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:21:53,921][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:21:54,419][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:21:54,920][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:21:55,423][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:21:55,924][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:21:56,424][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:21:56,924][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:21:57,425][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:21:57,926][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:21:58,426][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:21:58,928][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:21:59,428][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:21:59,929][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:22:00,429][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:22:00,930][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:22:01,429][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:22:01,930][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:22:02,430][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:22:02,930][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:22:03,429][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:22:03,928][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:22:04,427][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:22:04,926][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:22:05,425][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:22:05,925][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:22:06,424][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:22:06,923][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:22:07,424][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:22:07,923][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:22:08,423][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:22:08,923][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:22:09,423][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:22:09,924][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:22:10,423][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:22:10,922][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:22:11,421][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:22:11,920][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:22:12,419][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:22:12,917][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:22:13,417][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:22:13,916][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:22:14,415][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:22:14,918][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:22:15,420][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:22:15,924][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:22:16,425][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:22:16,926][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:22:17,426][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:22:17,927][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:22:18,429][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:22:18,928][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:22:19,428][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:22:19,926][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:22:20,425][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:22:20,924][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:22:21,422][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:22:21,920][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:22:22,419][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:22:22,916][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:22:23,414][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:22:23,912][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:22:24,410][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:22:24,910][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:22:25,409][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:22:25,906][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:22:26,405][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:22:26,904][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:22:27,404][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:22:27,904][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:22:28,404][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:22:28,905][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:22:29,405][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:22:29,904][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:22:30,403][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:22:30,904][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:22:31,405][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:22:31,905][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:22:32,404][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:22:32,903][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:22:33,403][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:22:33,902][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:22:34,402][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:22:34,901][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:22:35,401][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:22:35,900][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:22:36,398][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:22:36,897][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:22:37,396][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:22:37,894][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:22:38,392][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:22:38,891][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:22:39,390][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:22:39,889][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:22:40,387][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:22:40,885][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:22:41,384][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:22:41,883][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:22:42,380][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:22:42,879][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:22:43,379][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:22:43,878][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:22:44,377][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:22:44,876][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:22:45,375][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:22:45,873][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:22:46,372][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:22:46,872][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:22:47,372][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:22:47,873][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:22:48,375][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:22:48,876][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:22:49,375][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:22:49,875][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:22:50,375][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21747 tokens. [2026-03-26 07:22:51,008][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:22:51,742][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:22:51,744][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:22:51,746][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:22:52,587][__main__][INFO] - Iteration 512 took 1m 15s (10.15% Gen, 88.73% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 59m 23s. Estimated total time: 62h 47m 33s. Time estimates for 10 more iterations: 12m 33s, 100 more iterations: 2h 5m 35s, 500 more iterations: 10h 27m 55s. [2026-03-26 07:22:52,589][__main__][INFO] - Starting iteration 512. [2026-03-26 07:22:52,991][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:22:52,992][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:22:59,496][__main__][INFO] - Number of regex retries in iteration 512: 0 [2026-03-26 07:22:59,497][__main__][INFO] - agents played in iteration 512 are Alice, Bob [2026-03-26 07:23:00,438][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:23:00,484][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:23:00,484][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:23:00,485][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:23:01,093][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:23:01,549][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:23:02,053][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:23:02,552][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:23:03,052][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:23:03,552][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:23:04,052][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:23:04,552][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:23:05,051][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:23:05,554][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:23:06,056][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:23:06,558][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:23:07,059][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:23:07,560][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:23:08,062][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:23:08,563][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:23:09,067][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:23:09,567][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:23:10,067][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:23:10,566][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:23:11,066][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:23:11,566][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:23:12,066][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:23:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:23:13,066][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:23:13,566][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:23:14,065][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:23:14,563][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:23:15,062][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:23:15,560][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:23:16,060][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:23:16,559][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:23:17,057][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:23:17,557][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:23:18,058][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:23:18,558][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:23:19,057][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:23:19,556][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:23:20,055][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:23:20,555][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:23:21,055][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:23:21,555][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:23:22,055][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:23:22,555][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:23:23,055][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:23:23,556][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:23:24,057][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:23:24,557][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:23:25,059][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:23:25,558][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:23:26,057][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:23:26,555][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:23:27,055][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:23:27,554][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:23:28,053][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:23:28,552][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:23:29,051][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:23:29,550][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:23:30,049][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:23:30,547][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:23:31,046][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:23:31,545][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:23:32,044][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:23:32,543][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:23:33,041][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:23:33,541][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:23:34,041][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:23:34,541][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:23:35,042][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:23:35,542][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:23:36,041][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:23:36,540][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:23:37,039][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:23:37,540][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:23:38,045][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:23:38,547][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:23:39,050][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:23:39,553][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:23:40,055][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:23:40,556][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:23:41,059][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:23:41,558][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:23:42,058][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:23:42,557][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:23:43,057][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:23:43,556][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:23:44,054][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:23:44,554][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:23:45,053][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:23:45,555][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:23:46,055][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:23:46,553][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:23:47,051][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:23:47,550][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:23:48,050][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:23:48,551][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:23:49,050][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:23:49,549][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:23:50,047][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:23:50,547][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:23:51,047][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:23:51,546][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:23:52,045][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:23:52,544][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:23:53,044][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:23:53,545][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:23:54,045][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:23:54,547][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:23:55,048][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:23:55,549][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:23:56,050][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:23:56,551][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:23:57,079][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:23:57,578][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:23:58,077][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:23:58,576][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:23:59,074][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:23:59,573][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:24:00,071][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:24:00,570][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:24:01,070][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:24:01,569][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:24:02,068][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:24:02,567][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:24:03,066][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:24:03,564][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:24:04,063][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:24:04,560][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:24:05,059][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21744 tokens. [2026-03-26 07:24:05,694][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.52%, ΔTime: 00:01:04 [2026-03-26 07:24:06,436][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:24:06,439][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:24:06,440][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:24:07,088][__main__][INFO] - Iteration 513 took 1m 14s (8.78% Gen, 90.34% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 55m 27s. Estimated total time: 61h 44m 52s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 29s, 500 more iterations: 10h 17m 28s. [2026-03-26 07:24:07,090][__main__][INFO] - Starting iteration 513. [2026-03-26 07:24:07,491][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:24:07,491][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:24:14,767][__main__][INFO] - Number of regex retries in iteration 513: 0 [2026-03-26 07:24:14,768][__main__][INFO] - agents played in iteration 513 are Alice, Bob [2026-03-26 07:24:15,655][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:24:15,700][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:24:15,700][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:24:15,701][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:24:16,311][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:24:16,767][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:24:17,270][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:24:17,770][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:24:18,272][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:24:18,772][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:24:19,271][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:24:19,771][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:24:20,271][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:24:20,770][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:24:21,270][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:24:21,769][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:24:22,268][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:24:22,768][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:24:23,267][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:24:23,766][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:24:24,265][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:24:24,766][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:24:25,267][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:24:25,767][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:24:26,267][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:24:26,769][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:24:27,269][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:24:27,769][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:24:28,270][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:24:28,770][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:24:29,269][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:24:29,768][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:24:30,267][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:24:30,766][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:24:31,265][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:24:31,765][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:24:32,264][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:24:32,763][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:24:33,261][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:24:33,761][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:24:34,261][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:24:34,761][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:24:35,260][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:24:35,760][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:24:36,259][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:24:36,759][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:24:37,259][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:24:37,758][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:24:38,258][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:24:38,758][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:24:39,258][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:24:39,758][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:24:40,257][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:24:40,755][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:24:41,255][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:24:41,754][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:24:42,254][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:24:42,754][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:24:43,254][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:24:43,753][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:24:44,252][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:24:44,752][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:24:45,253][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:24:45,753][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:24:46,253][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:24:46,753][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:24:47,253][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:24:47,752][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:24:48,255][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:24:48,755][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:24:49,253][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:24:49,751][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:24:50,251][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:24:50,750][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:24:51,249][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:24:51,748][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:24:52,247][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:24:52,745][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:24:53,244][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:24:53,743][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:24:54,241][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:24:54,739][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:24:55,237][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:24:55,735][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:24:56,233][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:24:56,733][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:24:57,233][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:24:57,733][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:24:58,233][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:24:58,732][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:24:59,231][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:24:59,732][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:25:00,233][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:25:00,732][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:25:01,231][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:25:01,729][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:25:02,229][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:25:02,727][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:25:03,226][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:25:03,725][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:25:04,224][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:25:04,722][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:25:05,221][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:25:05,719][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:25:06,218][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:25:06,717][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:25:07,214][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:25:07,713][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:25:08,212][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:25:08,713][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:25:09,212][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:25:09,712][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:25:10,212][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:25:10,711][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:25:11,210][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:25:11,711][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:25:12,211][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:25:12,711][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:25:13,210][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:25:13,709][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:25:14,209][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:25:14,708][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:25:15,207][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:25:15,711][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:25:16,211][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:25:16,711][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:25:17,211][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:25:17,711][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:25:18,213][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:25:18,713][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:25:19,213][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:25:19,713][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:25:20,213][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21748 tokens. [2026-03-26 07:25:20,846][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:25:21,639][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:25:21,641][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:25:21,643][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:25:22,349][__main__][INFO] - Iteration 514 took 1m 14s (9.72% Gen, 89.34% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 32m 16s. Estimated total time: 62h 22m 56s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 45s, 500 more iterations: 10h 23m 49s. [2026-03-26 07:25:22,351][__main__][INFO] - Starting iteration 514. [2026-03-26 07:25:22,752][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:25:22,753][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:25:25,541][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:25:26,288][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:25:29,750][__main__][INFO] - Number of regex retries in iteration 514: 2 [2026-03-26 07:25:29,751][__main__][INFO] - agents played in iteration 514 are Alice, Bob [2026-03-26 07:25:30,682][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:25:30,727][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:25:30,727][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:25:30,728][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:25:31,334][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:25:31,788][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:25:32,292][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:25:32,790][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:25:33,290][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:25:33,788][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:25:34,287][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:25:34,786][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:25:35,286][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:25:35,787][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:25:36,288][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:25:36,788][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:25:37,287][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:25:37,787][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:25:38,287][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:25:38,788][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:25:39,288][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:25:39,788][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:25:40,287][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:25:40,787][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:25:41,285][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:25:41,785][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:25:42,284][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:25:42,785][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:25:43,286][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:25:43,788][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:25:44,288][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:25:44,788][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:25:45,288][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:25:45,789][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:25:46,290][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:25:46,791][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:25:47,291][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:25:47,790][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:25:48,291][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:25:48,790][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:25:49,289][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:25:49,788][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:25:50,287][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:25:50,786][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:25:51,285][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:25:51,783][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:25:52,281][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:25:52,779][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:25:53,278][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:25:53,776][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:25:54,275][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:25:54,773][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:25:55,272][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:25:55,772][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:25:56,272][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:25:56,771][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:25:57,271][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:25:57,772][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:25:58,272][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:25:58,771][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:25:59,272][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:25:59,770][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:26:00,271][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:26:00,770][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:26:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:26:01,768][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:26:02,268][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:26:02,767][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:26:03,265][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:26:03,764][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:26:04,263][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:26:04,764][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:26:05,264][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:26:05,763][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:26:06,262][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:26:06,761][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:26:07,259][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:26:07,759][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:26:08,258][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:26:08,759][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:26:09,259][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:26:09,759][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:26:10,259][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:26:10,758][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:26:11,258][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:26:11,756][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:26:12,256][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:26:12,755][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:26:13,254][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:26:13,753][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:26:14,253][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:26:14,751][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:26:15,250][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:26:15,752][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:26:16,253][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:26:16,755][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:26:17,255][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:26:17,756][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:26:18,258][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:26:18,760][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:26:19,262][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:26:19,761][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:26:20,260][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:26:20,758][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:26:21,256][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:26:21,754][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:26:22,252][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:26:22,751][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:26:23,251][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:26:23,750][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:26:24,249][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:26:24,747][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:26:25,246][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:26:25,745][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:26:26,245][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:26:26,745][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:26:27,244][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:26:27,744][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:26:28,244][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:26:28,746][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:26:29,246][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:26:29,746][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:26:30,247][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:26:30,747][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:26:31,247][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:26:31,746][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:26:32,245][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:26:32,745][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:26:33,246][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:26:33,745][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:26:34,245][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:26:34,745][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:26:35,244][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21744 tokens. [2026-03-26 07:26:35,879][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:26:36,630][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:26:36,632][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:26:36,634][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:26:37,294][__main__][INFO] - Iteration 515 took 1m 14s (9.39% Gen, 89.72% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 15m 11s. Estimated total time: 62h 7m 6s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 14s, 500 more iterations: 10h 21m 11s. [2026-03-26 07:26:37,296][__main__][INFO] - Starting iteration 515. [2026-03-26 07:26:37,696][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:26:37,696][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:26:39,395][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:26:41,669][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:26:44,685][__main__][INFO] - Number of regex retries in iteration 515: 2 [2026-03-26 07:26:44,686][__main__][INFO] - agents played in iteration 515 are Alice, Bob [2026-03-26 07:26:45,629][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:26:45,674][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:26:45,675][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:26:45,675][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:26:46,282][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:26:46,738][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:26:47,242][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:26:47,741][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:26:48,244][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:26:48,744][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:26:49,244][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:26:49,744][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:26:50,243][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:26:50,742][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:26:51,242][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:26:51,743][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:26:52,243][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:26:52,743][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:26:53,242][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:26:53,741][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:26:54,240][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:26:54,739][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:26:55,238][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:26:55,735][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:26:56,233][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:26:56,731][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:26:57,230][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:26:57,728][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:26:58,227][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:26:58,725][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:26:59,226][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:26:59,725][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:27:00,223][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:27:00,721][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:27:01,219][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:27:01,716][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:27:02,214][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:27:02,711][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:27:03,211][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:27:03,710][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:27:04,208][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:27:04,707][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:27:05,205][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:27:05,704][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:27:06,203][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:27:06,702][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:27:07,201][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:27:07,700][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:27:08,199][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:27:08,698][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:27:09,197][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:27:09,697][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:27:10,197][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:27:10,699][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:27:11,199][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:27:11,699][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:27:12,199][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:27:12,699][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:27:13,198][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:27:13,699][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:27:14,199][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:27:14,698][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:27:15,197][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:27:15,696][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:27:16,195][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:27:16,694][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:27:17,193][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:27:17,692][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:27:18,195][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:27:18,695][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:27:19,195][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:27:19,693][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:27:20,193][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:27:20,692][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:27:21,191][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:27:21,690][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:27:22,189][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:27:22,688][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:27:23,188][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:27:23,687][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:27:24,187][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:27:24,685][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:27:25,185][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:27:25,686][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:27:26,185][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:27:26,684][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:27:27,183][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:27:27,682][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:27:28,181][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:27:28,681][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:27:29,180][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:27:29,679][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:27:30,179][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:27:30,678][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:27:31,177][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:27:31,675][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:27:32,172][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:27:32,671][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:27:33,170][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:27:33,669][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:27:34,168][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:27:34,666][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:27:35,165][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:27:35,664][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:27:36,162][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:27:36,660][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:27:37,159][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:27:37,657][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:27:38,156][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:27:38,655][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:27:39,154][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:27:39,653][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:27:40,153][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:27:40,652][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:27:41,151][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:27:41,651][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:27:42,151][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:27:42,652][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:27:43,152][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:27:43,652][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:27:44,153][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:27:44,653][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:27:45,154][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:27:45,655][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:27:46,155][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:27:46,655][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:27:47,154][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:27:47,653][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:27:48,154][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:27:48,655][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:27:49,155][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:27:49,655][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:27:50,154][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 07:27:50,791][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 07:27:51,537][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:27:51,540][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:27:51,541][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:27:52,198][__main__][INFO] - Iteration 516 took 1m 14s (9.38% Gen, 89.73% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 11m 59s. Estimated total time: 62h 5m 9s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 51s. [2026-03-26 07:27:52,202][__main__][INFO] - Starting iteration 516. [2026-03-26 07:27:52,603][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:27:52,604][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:27:59,268][__main__][INFO] - Number of regex retries in iteration 516: 0 [2026-03-26 07:27:59,269][__main__][INFO] - agents played in iteration 516 are Alice, Bob [2026-03-26 07:28:00,146][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:28:00,191][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:28:00,192][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:28:00,193][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:28:00,797][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:28:01,251][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:28:01,754][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:28:02,253][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:28:02,752][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:28:03,251][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:28:03,750][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:28:04,249][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:28:04,749][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:28:05,252][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:28:05,752][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:28:06,252][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:28:06,752][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:28:07,252][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:28:07,752][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:28:08,253][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:28:08,753][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:28:09,253][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:28:09,753][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:28:10,252][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:28:10,752][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:28:11,251][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:28:11,752][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:28:12,251][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:28:12,751][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:28:13,250][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:28:13,749][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:28:14,249][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:28:14,748][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:28:15,248][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:28:15,750][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:28:16,251][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:28:16,750][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:28:17,249][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:28:17,749][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:28:18,250][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:28:18,750][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:28:19,249][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:28:19,748][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:28:20,247][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:28:20,746][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:28:21,247][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:28:21,747][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:28:22,248][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:28:22,748][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:28:23,247][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:28:23,747][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:28:24,247][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:28:24,747][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:28:25,248][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:28:25,748][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:28:26,248][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:28:26,747][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:28:27,248][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:28:27,748][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:28:28,249][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:28:28,748][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:28:29,251][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:28:29,752][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:28:30,253][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:28:30,754][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:28:31,255][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:28:31,755][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:28:32,256][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:28:32,758][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:28:33,257][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:28:33,756][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:28:34,255][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:28:34,755][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:28:35,253][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:28:35,753][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:28:36,252][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:28:36,753][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:28:37,252][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:28:37,751][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:28:38,251][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:28:38,752][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:28:39,252][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:28:39,751][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:28:40,252][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:28:40,752][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:28:41,252][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:28:41,751][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:28:42,251][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:28:42,751][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:28:43,252][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:28:43,751][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:28:44,251][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:28:44,751][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:28:45,250][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:28:45,749][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:28:46,248][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:28:46,747][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:28:47,247][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:28:47,746][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:28:48,246][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:28:48,746][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:28:49,247][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:28:49,746][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:28:50,246][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:28:50,746][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:28:51,246][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:28:51,744][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:28:52,243][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:28:52,742][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:28:53,240][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:28:53,739][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:28:54,238][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:28:54,737][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:28:55,237][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:28:55,735][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:28:56,234][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:28:56,732][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:28:57,231][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:28:57,731][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:28:58,232][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:28:58,733][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:28:59,231][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:28:59,730][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:29:00,231][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:29:00,730][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:29:01,230][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:29:01,732][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:29:02,233][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:29:02,734][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:29:03,234][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:29:03,732][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:29:04,232][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:29:04,733][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:29:05,365][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 07:29:06,112][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:29:06,114][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:29:06,116][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:29:06,769][__main__][INFO] - Iteration 517 took 1m 14s (8.99% Gen, 90.13% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 53m 54s. Estimated total time: 61h 48m 19s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 36s, 500 more iterations: 10h 18m 3s. [2026-03-26 07:29:06,771][__main__][INFO] - Starting iteration 517. [2026-03-26 07:29:07,171][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:29:07,172][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:29:14,101][__main__][INFO] - Number of regex retries in iteration 517: 0 [2026-03-26 07:29:14,102][__main__][INFO] - agents played in iteration 517 are Alice, Bob [2026-03-26 07:29:15,029][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:29:15,074][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:29:15,075][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:29:15,075][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:29:15,674][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:29:16,131][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:29:16,635][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:29:17,135][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:29:17,635][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:29:18,136][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:29:18,637][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:29:19,136][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:29:19,636][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:29:20,135][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:29:20,635][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:29:21,135][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:29:21,635][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:29:22,133][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:29:22,632][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:29:23,131][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:29:23,631][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:29:24,130][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:29:24,629][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:29:25,130][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:29:25,629][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:29:26,128][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:29:26,627][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:29:27,128][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:29:27,627][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:29:28,127][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:29:28,627][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:29:29,126][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:29:29,625][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:29:30,124][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:29:30,623][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:29:31,122][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:29:31,622][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:29:32,120][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:29:32,620][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:29:33,119][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:29:33,617][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:29:34,115][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:29:34,613][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:29:35,112][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:29:35,611][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:29:36,112][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:29:36,612][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:29:37,112][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:29:37,611][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:29:38,111][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:29:38,610][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:29:39,110][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:29:39,609][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:29:40,108][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:29:40,607][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:29:41,106][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:29:41,605][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:29:42,103][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:29:42,603][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:29:43,103][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:29:43,601][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:29:44,102][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:29:44,602][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:29:45,103][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:29:45,606][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:29:46,109][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:29:46,609][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:29:47,110][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:29:47,610][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:29:48,110][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:29:48,611][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:29:49,112][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:29:49,611][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:29:50,111][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:29:50,611][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:29:51,110][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:29:51,609][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:29:52,107][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:29:52,607][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:29:53,107][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:29:53,607][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:29:54,106][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:29:54,606][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:29:55,105][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:29:55,604][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:29:56,102][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:29:56,601][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:29:57,100][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:29:57,599][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:29:58,098][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:29:58,597][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:29:59,097][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:29:59,596][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:30:00,095][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:30:00,595][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:30:01,094][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:30:01,594][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:30:02,093][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:30:02,592][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:30:03,091][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:30:03,590][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:30:04,090][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:30:04,589][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:30:05,089][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:30:05,588][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:30:06,088][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:30:06,588][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:30:07,089][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:30:07,590][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:30:08,090][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:30:08,590][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:30:09,090][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:30:09,589][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:30:10,089][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:30:10,589][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:30:11,090][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:30:11,590][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:30:12,089][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:30:12,589][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:30:13,088][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:30:13,586][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:30:14,085][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:30:14,583][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:30:15,082][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:30:15,581][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:30:16,081][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:30:16,581][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:30:17,081][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:30:17,581][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:30:18,080][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:30:18,582][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:30:19,082][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:30:19,583][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21747 tokens. [2026-03-26 07:30:20,218][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.34%, ΔTime: 00:01:04 [2026-03-26 07:30:20,960][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:30:20,963][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:30:20,964][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:30:21,574][__main__][INFO] - Iteration 518 took 1m 14s (9.31% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 4m 31s. Estimated total time: 62h 0m 11s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 0s, 500 more iterations: 10h 20m 1s. [2026-03-26 07:30:21,576][__main__][INFO] - Starting iteration 518. [2026-03-26 07:30:21,979][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:30:21,979][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:30:22,554][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:30:29,042][__main__][INFO] - Number of regex retries in iteration 518: 1 [2026-03-26 07:30:29,043][__main__][INFO] - agents played in iteration 518 are Alice, Bob [2026-03-26 07:30:29,979][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:30:30,024][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:30:30,025][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:30:30,025][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:30:30,631][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:30:31,090][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:30:31,595][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:30:32,096][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:30:32,597][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:30:33,098][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:30:33,598][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:30:34,100][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:30:34,600][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:30:35,100][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:30:35,598][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:30:36,097][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:30:36,596][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:30:37,096][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:30:37,595][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:30:38,094][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:30:38,594][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:30:39,096][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:30:39,596][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:30:40,097][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:30:40,598][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:30:41,098][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:30:41,597][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:30:42,097][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:30:42,597][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:30:43,096][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:30:43,595][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:30:44,095][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:30:44,594][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:30:45,093][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:30:45,592][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:30:46,091][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:30:46,591][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:30:47,092][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:30:47,591][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:30:48,092][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:30:48,594][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:30:49,094][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:30:49,595][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:30:50,095][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:30:50,596][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:30:51,095][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:30:51,593][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:30:52,091][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:30:52,590][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:30:53,088][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:30:53,587][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:30:54,086][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:30:54,585][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:30:55,084][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:30:55,582][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:30:56,080][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:30:56,578][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:30:57,076][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:30:57,575][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:30:58,074][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:30:58,572][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:30:59,071][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:30:59,570][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:31:00,068][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:31:00,567][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:31:01,066][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:31:01,566][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:31:02,065][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:31:02,564][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:31:03,064][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:31:03,564][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:31:04,064][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:31:04,564][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:31:05,067][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:31:05,568][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:31:06,069][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:31:06,570][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:31:07,069][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:31:07,568][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:31:08,068][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:31:08,568][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:31:09,068][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:31:09,567][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:31:10,068][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:31:10,567][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:31:11,067][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:31:11,568][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:31:12,069][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:31:12,569][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:31:13,070][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:31:13,571][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:31:14,073][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:31:14,573][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:31:15,073][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:31:15,576][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:31:16,078][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:31:16,578][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:31:17,077][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:31:17,576][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:31:18,077][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:31:18,577][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:31:19,078][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:31:19,579][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:31:20,079][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:31:20,579][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:31:21,078][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:31:21,579][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:31:22,079][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:31:22,579][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:31:23,078][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:31:23,577][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:31:24,075][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:31:24,573][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:31:25,073][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:31:25,572][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:31:26,071][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:31:26,571][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:31:27,070][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:31:27,568][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:31:28,066][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:31:28,565][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:31:29,063][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:31:29,562][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:31:30,061][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:31:30,584][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:31:31,083][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:31:31,581][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:31:32,080][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:31:32,579][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:31:33,078][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:31:33,577][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:31:34,076][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:31:34,576][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 07:31:35,209][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:31:35,955][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:31:35,957][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:31:35,959][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:31:36,623][__main__][INFO] - Iteration 519 took 1m 14s (9.46% Gen, 89.65% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 15m 20s. Estimated total time: 62h 12m 14s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 2s. [2026-03-26 07:31:36,627][__main__][INFO] - Starting iteration 519. [2026-03-26 07:31:37,024][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:31:37,025][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:31:43,725][__main__][INFO] - Number of regex retries in iteration 519: 0 [2026-03-26 07:31:43,726][__main__][INFO] - agents played in iteration 519 are Alice, Bob [2026-03-26 07:31:44,613][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:31:44,658][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:31:44,658][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:31:44,659][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:31:45,264][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:31:45,720][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:31:46,225][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:31:46,726][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:31:47,226][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:31:47,727][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:31:48,229][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:31:48,730][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:31:49,229][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:31:49,730][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:31:50,229][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:31:50,728][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:31:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:31:51,725][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:31:52,223][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:31:52,721][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:31:53,219][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:31:53,717][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:31:54,216][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:31:54,715][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:31:55,214][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:31:55,713][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:31:56,210][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:31:56,708][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:31:57,206][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:31:57,708][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:31:58,208][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:31:58,708][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:31:59,207][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:31:59,707][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:32:00,207][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:32:00,707][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:32:01,208][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:32:01,710][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:32:02,211][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:32:02,712][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:32:03,211][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:32:03,712][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:32:04,212][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:32:04,712][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:32:05,213][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:32:05,713][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:32:06,213][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:32:06,713][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:32:07,213][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:32:07,714][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:32:08,215][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:32:08,715][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:32:09,215][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:32:09,714][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:32:10,213][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:32:10,711][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:32:11,211][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:32:11,711][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:32:12,210][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:32:12,710][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:32:13,210][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:32:13,729][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:32:14,231][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:32:14,732][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:32:15,234][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:32:15,734][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:32:16,235][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:32:16,735][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:32:17,237][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:32:17,737][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:32:18,239][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:32:18,737][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:32:19,236][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:32:19,733][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:32:20,232][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:32:20,731][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:32:21,230][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:32:21,730][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:32:22,229][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:32:22,726][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:32:23,225][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:32:23,723][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:32:24,221][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:32:24,719][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:32:25,219][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:32:25,717][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:32:26,215][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:32:26,713][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:32:27,211][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:32:27,709][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:32:28,207][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:32:28,707][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:32:29,205][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:32:29,705][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:32:30,205][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:32:30,704][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:32:31,203][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:32:31,702][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:32:32,201][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:32:32,701][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:32:33,200][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:32:33,700][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:32:34,199][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:32:34,699][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:32:35,199][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:32:35,699][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:32:36,198][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:32:36,699][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:32:37,199][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:32:37,699][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:32:38,199][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:32:38,698][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:32:39,198][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:32:39,697][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:32:40,196][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:32:40,696][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:32:41,197][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:32:41,695][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:32:42,194][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:32:42,693][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:32:43,193][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:32:43,692][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:32:44,192][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:32:44,691][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:32:45,191][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:32:45,695][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:32:46,198][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:32:46,698][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:32:47,200][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:32:47,699][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:32:48,202][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:32:48,703][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:32:49,203][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 07:32:49,868][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:32:50,604][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:32:50,607][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:32:50,608][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:32:51,211][__main__][INFO] - Iteration 520 took 1m 14s (9.03% Gen, 90.15% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 51m 12s. Estimated total time: 61h 49m 21s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 38s, 500 more iterations: 10h 18m 13s. [2026-03-26 07:32:51,213][__main__][INFO] - Starting iteration 520. [2026-03-26 07:32:51,617][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:32:51,618][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:32:53,861][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:32:58,452][__main__][INFO] - Number of regex retries in iteration 520: 1 [2026-03-26 07:32:58,452][__main__][INFO] - agents played in iteration 520 are Alice, Bob [2026-03-26 07:32:59,332][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:32:59,377][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:32:59,377][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:32:59,378][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:32:59,983][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:33:00,439][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:33:00,944][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:33:01,448][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:33:01,948][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:33:02,448][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:33:02,950][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:33:03,452][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:33:03,954][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:33:04,455][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:33:04,955][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:33:05,455][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:33:05,955][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:33:06,456][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:33:06,957][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:33:07,457][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:33:07,957][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:33:08,456][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:33:08,955][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:33:09,454][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:33:09,954][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:33:10,453][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:33:10,952][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:33:11,450][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:33:11,949][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:33:12,450][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:33:12,949][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:33:13,448][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:33:13,948][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:33:14,447][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:33:14,946][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:33:15,444][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:33:15,942][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:33:16,443][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:33:16,943][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:33:17,443][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:33:17,944][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:33:18,447][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:33:18,947][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:33:19,446][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:33:19,948][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:33:20,448][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:33:20,948][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:33:21,447][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:33:21,948][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:33:22,447][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:33:22,945][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:33:23,443][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:33:23,943][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:33:24,439][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:33:24,936][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:33:25,434][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:33:25,932][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:33:26,429][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:33:26,925][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:33:27,422][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:33:27,919][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:33:28,418][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:33:28,917][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:33:29,417][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:33:29,917][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:33:30,416][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:33:30,915][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:33:31,414][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:33:31,914][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:33:32,412][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:33:32,912][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:33:33,411][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:33:33,910][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:33:34,409][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:33:34,908][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:33:35,407][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:33:35,906][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:33:36,407][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:33:36,908][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:33:37,408][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:33:37,907][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:33:38,406][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:33:38,905][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:33:39,404][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:33:39,903][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:33:40,402][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:33:40,900][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:33:41,398][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:33:41,897][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:33:42,396][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:33:42,894][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:33:43,393][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:33:43,892][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:33:44,391][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:33:44,890][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:33:45,390][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:33:45,888][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:33:46,387][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:33:46,885][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:33:47,383][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:33:47,882][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:33:48,384][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:33:48,886][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:33:49,386][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:33:49,886][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:33:50,387][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:33:50,886][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:33:51,388][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:33:51,889][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:33:52,389][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:33:52,888][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:33:53,387][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:33:53,886][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:33:54,384][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:33:54,883][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:33:55,382][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:33:55,881][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:33:56,377][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:33:56,875][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:33:57,373][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:33:57,870][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:33:58,367][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:33:58,863][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:33:59,360][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:33:59,858][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:34:00,360][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:34:00,861][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:34:01,361][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:34:01,861][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:34:02,360][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:34:02,861][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:34:03,361][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:34:03,861][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:34:04,594][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:34:05,279][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:34:05,282][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:34:05,283][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:34:06,181][__main__][INFO] - Iteration 521 took 1m 14s (9.17% Gen, 89.63% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 8m 48s. Estimated total time: 62h 8m 12s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 16s, 500 more iterations: 10h 21m 22s. [2026-03-26 07:34:06,183][__main__][INFO] - Starting iteration 521. [2026-03-26 07:34:06,582][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:34:06,583][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:34:13,572][__main__][INFO] - Number of regex retries in iteration 521: 0 [2026-03-26 07:34:13,573][__main__][INFO] - agents played in iteration 521 are Alice, Bob [2026-03-26 07:34:14,471][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:34:14,516][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:34:14,517][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:34:14,517][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:34:15,123][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:34:15,580][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:34:16,086][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:34:16,585][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:34:17,086][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:34:17,586][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:34:18,087][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:34:18,588][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:34:19,088][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:34:19,586][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:34:20,087][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:34:20,587][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:34:21,087][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:34:21,587][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:34:22,086][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:34:22,585][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:34:23,084][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:34:23,582][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:34:24,081][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:34:24,580][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:34:25,080][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:34:25,579][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:34:26,078][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:34:26,576][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:34:27,075][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:34:27,576][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:34:28,077][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:34:28,577][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:34:29,077][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:34:29,576][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:34:30,075][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:34:30,575][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:34:31,075][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:34:31,577][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:34:32,078][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:34:32,578][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:34:33,079][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:34:33,579][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:34:34,079][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:34:34,579][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:34:35,080][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:34:35,579][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:34:36,077][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:34:36,576][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:34:37,075][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:34:37,574][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:34:38,074][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:34:38,573][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:34:39,072][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:34:39,573][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:34:40,072][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:34:40,570][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:34:41,070][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:34:41,569][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:34:42,069][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:34:42,569][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:34:43,068][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:34:43,568][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:34:44,067][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:34:44,566][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:34:45,064][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:34:45,563][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:34:46,063][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:34:46,562][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:34:47,062][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:34:47,562][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:34:48,063][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:34:48,564][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:34:49,065][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:34:49,566][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:34:50,067][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:34:50,568][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:34:51,068][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:34:51,567][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:34:52,065][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:34:52,565][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:34:53,065][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:34:53,564][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:34:54,063][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:34:54,562][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:34:55,061][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:34:55,560][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:34:56,059][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:34:56,560][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:34:57,060][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:34:57,560][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:34:58,059][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:34:58,558][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:34:59,056][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:34:59,556][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:35:00,055][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:35:00,556][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:35:01,056][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:35:01,555][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:35:02,055][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:35:02,554][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:35:03,054][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:35:03,555][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:35:04,056][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:35:04,557][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:35:05,057][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:35:05,558][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:35:06,057][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:35:06,558][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:35:07,059][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:35:07,559][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:35:08,058][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:35:08,557][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:35:09,055][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:35:09,554][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:35:10,053][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:35:10,552][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:35:11,051][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:35:11,550][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:35:12,049][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:35:12,549][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:35:13,048][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:35:13,546][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:35:14,046][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:35:14,545][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:35:15,044][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:35:15,543][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:35:16,042][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:35:16,540][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:35:17,039][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:35:17,538][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:35:18,037][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:35:18,538][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:35:19,037][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:35:19,672][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:35:20,402][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:35:20,405][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:35:20,406][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:35:21,199][__main__][INFO] - Iteration 522 took 1m 14s (9.37% Gen, 89.57% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 10m 13s. Estimated total time: 62h 10m 52s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 21s, 500 more iterations: 10h 21m 48s. [2026-03-26 07:35:21,201][__main__][INFO] - Starting iteration 522. [2026-03-26 07:35:21,601][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:35:21,602][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:35:28,484][__main__][INFO] - Number of regex retries in iteration 522: 0 [2026-03-26 07:35:28,485][__main__][INFO] - agents played in iteration 522 are Alice, Bob [2026-03-26 07:35:29,367][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:35:29,412][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:35:29,412][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:35:29,413][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:35:30,014][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:35:30,467][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:35:30,969][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:35:31,468][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:35:31,966][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:35:32,465][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:35:32,964][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:35:33,462][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:35:33,961][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:35:34,461][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:35:34,961][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:35:35,459][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:35:35,958][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:35:36,458][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:35:36,957][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:35:37,457][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:35:37,957][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:35:38,458][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:35:38,958][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:35:39,457][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:35:39,959][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:35:40,458][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:35:40,958][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:35:41,458][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:35:41,958][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:35:42,457][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:35:42,956][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:35:43,457][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:35:43,958][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:35:44,458][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:35:44,957][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:35:45,460][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:35:45,962][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:35:46,461][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:35:46,960][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:35:47,459][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:35:47,960][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:35:48,461][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:35:48,962][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:35:49,460][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:35:49,960][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:35:50,460][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:35:50,960][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:35:51,460][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:35:51,960][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:35:52,460][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:35:52,959][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:35:53,459][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:35:53,960][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:35:54,459][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:35:54,958][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:35:55,457][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:35:55,956][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:35:56,456][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:35:56,955][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:35:57,455][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:35:57,955][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:35:58,454][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:35:58,953][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:35:59,452][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:35:59,952][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:36:00,452][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:36:00,951][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:36:01,452][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:36:01,951][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:36:02,449][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:36:02,948][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:36:03,446][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:36:03,945][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:36:04,443][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:36:04,940][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:36:05,438][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:36:05,939][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:36:06,440][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:36:06,940][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:36:07,440][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:36:07,939][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:36:08,438][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:36:08,937][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:36:09,436][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:36:09,935][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:36:10,435][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:36:10,934][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:36:11,435][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:36:11,935][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:36:12,435][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:36:12,935][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:36:13,436][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:36:13,935][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:36:14,435][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:36:14,935][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:36:15,434][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:36:15,934][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:36:16,435][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:36:16,935][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:36:17,435][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:36:17,935][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:36:18,436][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:36:18,935][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:36:19,434][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:36:19,933][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:36:20,432][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:36:20,931][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:36:21,431][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:36:21,930][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:36:22,431][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:36:22,930][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:36:23,430][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:36:23,930][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:36:24,429][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:36:24,928][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:36:25,429][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:36:25,929][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:36:26,428][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:36:26,929][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:36:27,429][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:36:27,929][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:36:28,428][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:36:28,927][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:36:29,426][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:36:29,925][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:36:30,423][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:36:30,920][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:36:31,418][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:36:31,917][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:36:32,415][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:36:32,914][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:36:33,412][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:36:33,910][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 07:36:34,538][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-26 07:36:35,276][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:36:35,279][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:36:35,280][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:36:35,933][__main__][INFO] - Iteration 523 took 1m 14s (9.26% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 54m 44s. Estimated total time: 61h 56m 38s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 53s, 500 more iterations: 10h 19m 26s. [2026-03-26 07:36:35,936][__main__][INFO] - Starting iteration 523. [2026-03-26 07:36:36,335][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:36:36,336][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:36:43,296][__main__][INFO] - Number of regex retries in iteration 523: 0 [2026-03-26 07:36:43,297][__main__][INFO] - agents played in iteration 523 are Alice, Bob [2026-03-26 07:36:44,182][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:36:44,227][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:36:44,228][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:36:44,228][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:36:44,829][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:36:45,284][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:36:45,787][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:36:46,286][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:36:46,784][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:36:47,283][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:36:47,781][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:36:48,282][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:36:48,781][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:36:49,281][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:36:49,780][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:36:50,278][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:36:50,777][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:36:51,276][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:36:51,776][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:36:52,275][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:36:52,774][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:36:53,273][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:36:53,772][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:36:54,269][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:36:54,768][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:36:55,266][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:36:55,764][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:36:56,262][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:36:56,761][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:36:57,261][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:36:57,760][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:36:58,259][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:36:58,758][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:36:59,257][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:36:59,757][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:37:00,257][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:37:00,755][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:37:01,254][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:37:01,753][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:37:02,252][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:37:02,752][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:37:03,251][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:37:03,750][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:37:04,249][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:37:04,750][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:37:05,250][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:37:05,749][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:37:06,250][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:37:06,750][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:37:07,250][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:37:07,750][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:37:08,249][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:37:08,749][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:37:09,250][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:37:09,751][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:37:10,251][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:37:10,751][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:37:11,251][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:37:11,750][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:37:12,253][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:37:12,753][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:37:13,254][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:37:13,753][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:37:14,253][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:37:14,753][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:37:15,253][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:37:15,760][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:37:16,261][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:37:16,760][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:37:17,260][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:37:17,759][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:37:18,260][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:37:18,759][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:37:19,257][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:37:19,757][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:37:20,256][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:37:20,755][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:37:21,255][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:37:21,754][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:37:22,253][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:37:22,751][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:37:23,251][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:37:23,750][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:37:24,250][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:37:24,750][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:37:25,250][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:37:25,749][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:37:26,247][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:37:26,744][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:37:27,243][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:37:27,741][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:37:28,239][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:37:28,737][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:37:29,236][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:37:29,736][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:37:30,235][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:37:30,734][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:37:31,234][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:37:31,733][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:37:32,233][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:37:32,732][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:37:33,231][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:37:33,730][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:37:34,229][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:37:34,728][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:37:35,227][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:37:35,725][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:37:36,224][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:37:36,723][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:37:37,221][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:37:37,720][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:37:38,219][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:37:38,718][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:37:39,216][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:37:39,714][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:37:40,213][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:37:40,712][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:37:41,211][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:37:41,712][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:37:42,211][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:37:42,713][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:37:43,213][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:37:43,712][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:37:44,211][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:37:44,711][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:37:45,213][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:37:45,713][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:37:46,213][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:37:46,713][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:37:47,213][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:37:47,713][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:37:48,216][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:37:48,718][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21750 tokens. [2026-03-26 07:37:49,366][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.23%, ΔTime: 00:01:04 [2026-03-26 07:37:50,104][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:37:50,106][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:37:50,108][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:37:50,761][__main__][INFO] - Iteration 524 took 1m 14s (9.35% Gen, 89.77% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 58m 10s. Estimated total time: 62h 1m 19s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 2s, 500 more iterations: 10h 20m 13s. [2026-03-26 07:37:50,763][__main__][INFO] - Starting iteration 524. [2026-03-26 07:37:51,162][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:37:51,163][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:37:51,750][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:37:52,416][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:37:58,023][__main__][INFO] - Number of regex retries in iteration 524: 2 [2026-03-26 07:37:58,024][__main__][INFO] - agents played in iteration 524 are Alice, Bob [2026-03-26 07:37:58,904][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:37:58,950][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:37:58,951][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:37:58,951][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:37:59,559][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:38:00,015][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:38:00,520][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:38:01,020][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:38:01,519][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:38:02,019][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:38:02,524][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:38:03,024][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:38:03,524][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:38:04,024][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:38:04,524][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:38:05,024][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:38:05,523][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:38:06,029][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:38:06,530][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:38:07,033][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:38:07,534][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:38:08,037][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:38:08,538][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:38:09,039][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:38:09,540][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:38:10,045][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:38:10,547][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:38:11,049][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:38:11,551][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:38:12,052][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:38:12,553][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:38:13,053][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:38:13,558][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:38:14,060][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:38:14,561][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:38:15,061][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:38:15,562][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:38:16,063][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:38:16,564][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:38:17,065][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:38:17,573][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:38:18,074][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:38:18,578][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:38:19,081][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:38:19,583][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:38:20,087][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:38:20,590][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:38:21,098][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:38:21,601][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:38:22,104][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:38:22,606][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:38:23,108][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:38:23,610][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:38:24,112][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:38:24,612][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:38:25,114][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:38:25,614][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:38:26,115][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:38:26,616][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:38:27,116][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:38:27,614][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:38:28,114][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:38:28,614][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:38:29,114][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:38:29,613][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:38:30,112][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:38:30,611][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:38:31,112][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:38:31,612][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:38:32,113][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:38:32,613][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:38:33,113][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:38:33,614][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:38:34,114][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:38:34,615][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:38:35,116][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:38:35,617][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:38:36,116][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:38:36,615][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:38:37,114][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:38:37,613][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:38:38,112][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:38:38,612][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:38:39,111][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:38:39,611][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:38:40,112][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:38:40,613][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:38:41,113][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:38:41,613][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:38:42,113][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:38:42,615][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:38:43,115][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:38:43,615][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:38:44,116][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:38:44,615][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:38:45,114][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:38:45,616][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:38:46,117][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:38:46,615][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:38:47,114][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:38:47,613][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:38:48,114][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:38:48,617][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:38:49,118][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:38:49,618][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:38:50,118][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:38:50,618][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:38:51,117][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:38:51,618][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:38:52,119][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:38:52,619][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:38:53,118][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:38:53,618][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:38:54,117][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:38:54,617][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:38:55,117][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:38:55,617][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:38:56,116][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:38:56,616][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:38:57,114][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:38:57,613][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:38:58,111][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:38:58,610][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:38:59,109][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:38:59,609][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:39:00,107][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:39:00,605][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:39:01,103][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:39:01,601][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:39:02,099][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:39:02,596][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:39:03,095][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:39:03,593][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21753 tokens. [2026-03-26 07:39:04,232][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:39:04,971][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:39:04,973][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:39:04,975][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:39:05,684][__main__][INFO] - Iteration 525 took 1m 14s (9.21% Gen, 89.84% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 51h 1m 43s. Estimated total time: 62h 6m 6s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 12s, 500 more iterations: 10h 21m 1s. [2026-03-26 07:39:05,686][__main__][INFO] - Starting iteration 525. [2026-03-26 07:39:06,085][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:39:06,086][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:39:13,226][__main__][INFO] - Number of regex retries in iteration 525: 0 [2026-03-26 07:39:13,227][__main__][INFO] - agents played in iteration 525 are Alice, Bob [2026-03-26 07:39:14,110][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:39:14,155][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:39:14,156][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:39:14,156][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:39:14,762][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:39:15,213][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:39:15,717][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:39:16,214][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:39:16,711][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:39:17,208][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:39:17,706][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:39:18,205][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:39:18,703][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:39:19,204][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:39:19,703][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:39:20,203][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:39:20,701][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:39:21,201][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:39:21,701][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:39:22,201][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:39:22,700][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:39:23,198][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:39:23,697][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:39:24,196][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:39:24,695][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:39:25,193][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:39:25,692][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:39:26,190][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:39:26,689][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:39:27,191][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:39:27,692][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:39:28,192][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:39:28,691][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:39:29,192][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:39:29,692][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:39:30,193][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:39:30,693][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:39:31,193][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:39:31,693][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:39:32,192][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:39:32,691][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:39:33,191][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:39:33,692][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:39:34,192][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:39:34,692][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:39:35,192][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:39:35,691][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:39:36,191][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:39:36,690][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:39:37,191][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:39:37,691][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:39:38,191][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:39:38,691][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:39:39,190][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:39:39,689][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:39:40,188][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:39:40,687][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:39:41,188][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:39:41,687][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:39:42,187][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:39:42,687][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:39:43,186][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:39:43,685][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:39:44,185][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:39:44,685][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:39:45,184][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:39:45,685][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:39:46,184][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:39:46,684][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:39:47,180][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:39:47,678][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:39:48,177][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:39:48,676][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:39:49,174][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:39:49,672][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:39:50,169][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:39:50,666][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:39:51,165][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:39:51,664][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:39:52,163][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:39:52,663][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:39:53,162][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:39:53,662][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:39:54,162][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:39:54,661][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:39:55,160][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:39:55,658][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:39:56,157][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:39:56,656][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:39:57,154][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:39:57,653][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:39:58,152][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:39:58,652][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:39:59,153][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:39:59,653][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:40:00,153][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:40:00,655][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:40:01,155][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:40:01,654][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:40:02,155][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:40:02,655][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:40:03,156][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:40:03,656][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:40:04,156][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:40:04,657][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:40:05,157][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:40:05,656][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:40:06,156][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:40:06,657][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:40:07,158][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:40:07,658][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:40:08,158][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:40:08,658][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:40:09,158][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:40:09,658][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:40:10,157][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:40:10,657][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:40:11,158][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:40:11,657][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:40:12,157][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:40:12,657][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:40:13,156][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:40:13,657][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:40:14,158][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:40:14,658][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:40:15,157][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:40:15,658][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:40:16,157][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:40:16,657][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:40:17,158][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:40:17,658][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:40:18,159][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:40:18,660][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:40:19,293][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.25%, ΔTime: 00:01:04 [2026-03-26 07:40:20,075][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:40:20,077][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:40:20,079][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:40:20,792][__main__][INFO] - Iteration 526 took 1m 14s (9.56% Gen, 89.49% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 9m 42s. Estimated total time: 62h 15m 21s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 30s, 500 more iterations: 10h 22m 33s. [2026-03-26 07:40:20,794][__main__][INFO] - Starting iteration 526. [2026-03-26 07:40:21,194][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:40:21,195][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:40:23,354][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:40:25,782][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:40:28,057][__main__][INFO] - Number of regex retries in iteration 526: 2 [2026-03-26 07:40:28,058][__main__][INFO] - agents played in iteration 526 are Alice, Bob [2026-03-26 07:40:28,938][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:40:28,983][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:40:28,984][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:40:28,985][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:40:29,587][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:40:30,041][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:40:30,545][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:40:31,044][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:40:31,542][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:40:32,041][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:40:32,539][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:40:33,038][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:40:33,537][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:40:34,039][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:40:34,540][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:40:35,040][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:40:35,539][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:40:36,039][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:40:36,539][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:40:37,039][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:40:37,540][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:40:38,040][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:40:38,540][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:40:39,040][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:40:39,540][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:40:40,039][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:40:40,539][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:40:41,041][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:40:41,541][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:40:42,041][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:40:42,541][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:40:43,040][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:40:43,540][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:40:44,039][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:40:44,540][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:40:45,040][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:40:45,540][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:40:46,040][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:40:46,539][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:40:47,038][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:40:47,537][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:40:48,038][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:40:48,540][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:40:49,039][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:40:49,538][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:40:50,037][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:40:50,535][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:40:51,033][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:40:51,533][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:40:52,031][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:40:52,530][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:40:53,028][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:40:53,525][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:40:54,024][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:40:54,522][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:40:55,021][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:40:55,520][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:40:56,019][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:40:56,517][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:40:57,016][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:40:57,515][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:40:58,012][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:40:58,510][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:40:59,008][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:40:59,506][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:41:00,004][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:41:00,502][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:41:01,000][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:41:01,497][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:41:01,995][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:41:02,494][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:41:02,992][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:41:03,490][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:41:03,990][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:41:04,490][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:41:04,989][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:41:05,488][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:41:05,987][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:41:06,486][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:41:06,985][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:41:07,483][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:41:07,982][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:41:08,481][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:41:08,981][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:41:09,480][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:41:09,979][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:41:10,478][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:41:10,977][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:41:11,476][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:41:11,975][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:41:12,474][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:41:12,974][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:41:13,473][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:41:13,973][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:41:14,472][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:41:14,972][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:41:15,472][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:41:15,972][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:41:16,472][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:41:16,972][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:41:17,471][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:41:17,971][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:41:18,471][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:41:18,972][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:41:19,471][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:41:19,971][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:41:20,471][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:41:20,969][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:41:21,468][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:41:21,965][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:41:22,463][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:41:22,961][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:41:23,460][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:41:23,959][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:41:24,457][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:41:24,955][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:41:25,454][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:41:25,954][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:41:26,453][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:41:26,952][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:41:27,451][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:41:27,950][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:41:28,449][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:41:28,947][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:41:29,446][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:41:29,943][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:41:30,441][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:41:30,939][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:41:31,437][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:41:31,934][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:41:32,432][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:41:32,929][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:41:33,426][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:41:34,070][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 07:41:34,810][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:41:34,812][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:41:34,814][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:41:35,542][__main__][INFO] - Iteration 527 took 1m 14s (9.23% Gen, 89.79% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 50m 30s. Estimated total time: 61h 57m 24s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 54s, 500 more iterations: 10h 19m 34s. [2026-03-26 07:41:35,544][__main__][INFO] - Starting iteration 527. [2026-03-26 07:41:35,944][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:41:35,945][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:41:36,531][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:41:38,652][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:41:42,462][__main__][INFO] - Number of regex retries in iteration 527: 2 [2026-03-26 07:41:42,463][__main__][INFO] - agents played in iteration 527 are Alice, Bob [2026-03-26 07:41:43,338][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:41:43,383][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:41:43,384][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:41:43,384][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:41:43,999][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:41:44,453][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:41:44,958][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:41:45,460][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:41:45,963][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:41:46,464][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:41:46,964][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:41:47,463][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:41:47,964][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:41:48,466][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:41:48,964][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:41:49,463][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:41:49,961][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:41:50,459][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:41:50,958][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:41:51,458][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:41:51,957][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:41:52,459][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:41:52,959][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:41:53,459][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:41:53,960][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:41:54,460][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:41:54,961][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:41:55,462][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:41:55,963][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:41:56,462][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:41:56,961][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:41:57,459][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:41:57,959][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:41:58,459][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:41:58,957][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:41:59,457][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:41:59,956][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:42:00,456][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:42:00,957][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:42:01,458][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:42:01,959][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:42:02,459][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:42:02,959][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:42:03,459][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:42:03,958][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:42:04,457][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:42:04,957][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:42:05,455][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:42:05,954][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:42:06,453][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:42:06,952][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:42:07,451][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:42:07,950][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:42:08,451][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:42:08,952][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:42:09,452][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:42:09,953][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:42:10,452][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:42:10,951][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:42:11,453][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:42:11,952][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:42:12,454][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:42:12,954][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:42:13,454][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:42:13,955][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:42:14,456][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:42:14,957][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:42:15,458][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:42:15,958][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:42:16,457][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:42:16,957][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:42:17,456][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:42:17,957][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:42:18,458][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:42:18,958][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:42:19,458][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:42:19,958][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:42:20,457][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:42:20,955][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:42:21,454][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:42:21,954][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:42:22,454][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:42:22,952][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:42:23,452][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:42:23,951][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:42:24,452][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:42:24,952][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:42:25,454][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:42:25,954][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:42:26,455][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:42:26,955][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:42:27,456][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:42:27,956][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:42:28,455][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:42:28,954][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:42:29,454][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:42:29,953][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:42:30,452][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:42:30,951][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:42:31,451][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:42:31,949][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:42:32,448][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:42:32,947][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:42:33,447][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:42:33,947][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:42:34,447][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:42:34,946][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:42:35,445][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:42:35,945][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:42:36,442][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:42:36,941][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:42:37,440][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:42:37,938][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:42:38,438][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:42:38,937][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:42:39,435][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:42:39,932][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:42:40,432][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:42:40,930][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:42:41,431][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:42:41,931][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:42:42,430][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:42:42,930][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:42:43,428][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:42:43,927][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:42:44,426][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:42:44,925][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:42:45,425][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:42:45,924][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:42:46,423][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:42:46,923][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:42:47,423][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:42:47,923][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21760 tokens. [2026-03-26 07:42:48,559][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:42:49,296][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:42:49,298][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:42:49,300][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:42:49,935][__main__][INFO] - Iteration 528 took 1m 13s (8.81% Gen, 90.33% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 31m 27s. Estimated total time: 61h 39m 34s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 19s, 500 more iterations: 10h 16m 35s. [2026-03-26 07:42:49,938][__main__][INFO] - Starting iteration 528. [2026-03-26 07:42:50,343][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:42:50,344][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:42:50,932][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:42:56,895][__main__][INFO] - Number of regex retries in iteration 528: 1 [2026-03-26 07:42:56,896][__main__][INFO] - agents played in iteration 528 are Alice, Bob [2026-03-26 07:42:57,778][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:42:57,823][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:42:57,823][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:42:57,824][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:42:58,430][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:42:58,884][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:42:59,388][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:42:59,888][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:43:00,386][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:43:00,884][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:43:01,383][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:43:01,882][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:43:02,381][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:43:02,882][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:43:03,381][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:43:03,882][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:43:04,381][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:43:04,880][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:43:05,380][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:43:05,880][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:43:06,379][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:43:06,879][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:43:07,378][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:43:07,877][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:43:08,376][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:43:08,875][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:43:09,374][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:43:09,874][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:43:10,374][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:43:10,874][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:43:11,374][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:43:11,873][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:43:12,373][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:43:12,873][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:43:13,373][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:43:13,874][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:43:14,373][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:43:14,871][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:43:15,372][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:43:15,874][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:43:16,374][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:43:16,873][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:43:17,372][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:43:17,872][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:43:18,371][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:43:18,875][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:43:19,376][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:43:19,877][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:43:20,378][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:43:20,879][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:43:21,380][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:43:21,880][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:43:22,381][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:43:22,880][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:43:23,380][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:43:23,879][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:43:24,378][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:43:24,878][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:43:25,378][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:43:25,878][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:43:26,377][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:43:26,878][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:43:27,378][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:43:27,879][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:43:28,379][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:43:28,876][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:43:29,377][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:43:29,877][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:43:30,377][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:43:30,877][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:43:31,376][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:43:31,874][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:43:32,373][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:43:32,872][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:43:33,372][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:43:33,873][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:43:34,373][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:43:34,872][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:43:35,371][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:43:35,871][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:43:36,370][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:43:36,871][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:43:37,371][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:43:37,871][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:43:38,371][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:43:38,870][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:43:39,369][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:43:39,868][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:43:40,367][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:43:40,867][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:43:41,366][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:43:41,866][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:43:42,365][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:43:42,863][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:43:43,361][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:43:43,860][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:43:44,358][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:43:44,858][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:43:45,358][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:43:45,857][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:43:46,355][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:43:46,855][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:43:47,354][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:43:47,854][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:43:48,356][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:43:48,857][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:43:49,356][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:43:49,856][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:43:50,356][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:43:50,857][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:43:51,358][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:43:51,859][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:43:52,360][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:43:52,861][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:43:53,361][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:43:53,861][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:43:54,360][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:43:54,859][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:43:55,359][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:43:55,859][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:43:56,358][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:43:56,856][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:43:57,355][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:43:57,854][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:43:58,353][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:43:58,854][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:43:59,354][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:43:59,855][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:44:00,355][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:44:00,855][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:44:01,356][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:44:01,856][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:44:02,355][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21727 tokens. [2026-03-26 07:44:02,994][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 07:44:03,731][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:44:03,734][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:44:03,735][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:44:04,382][__main__][INFO] - Iteration 529 took 1m 14s (8.85% Gen, 90.27% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 32m 37s. Estimated total time: 61h 42m 0s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 24s, 500 more iterations: 10h 17m 0s. [2026-03-26 07:44:04,384][__main__][INFO] - Starting iteration 529. [2026-03-26 07:44:04,785][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:44:04,786][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:44:11,689][__main__][INFO] - Number of regex retries in iteration 529: 0 [2026-03-26 07:44:11,690][__main__][INFO] - agents played in iteration 529 are Alice, Bob [2026-03-26 07:44:12,565][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:44:12,610][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:44:12,611][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:44:12,611][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:44:13,212][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:44:13,667][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:44:14,171][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:44:14,671][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:44:15,171][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:44:15,671][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:44:16,171][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:44:16,670][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:44:17,169][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:44:17,668][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:44:18,169][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:44:18,669][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:44:19,169][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:44:19,668][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:44:20,166][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:44:20,665][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:44:21,164][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:44:21,663][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:44:22,162][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:44:22,661][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:44:23,161][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:44:23,661][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:44:24,159][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:44:24,659][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:44:25,159][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:44:25,659][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:44:26,159][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:44:26,658][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:44:27,158][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:44:27,657][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:44:28,157][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:44:28,656][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:44:29,154][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:44:29,651][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:44:30,149][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:44:30,645][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:44:31,144][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:44:31,641][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:44:32,138][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:44:32,636][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:44:33,134][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:44:33,635][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:44:34,134][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:44:34,633][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:44:35,132][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:44:35,631][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:44:36,129][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:44:36,627][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:44:37,126][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:44:37,626][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:44:38,125][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:44:38,625][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:44:39,125][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:44:39,623][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:44:40,122][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:44:40,622][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:44:41,120][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:44:41,623][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:44:42,123][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:44:42,624][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:44:43,124][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:44:43,624][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:44:44,124][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:44:44,623][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:44:45,124][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:44:45,626][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:44:46,126][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:44:46,625][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:44:47,124][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:44:47,623][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:44:48,121][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:44:48,622][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:44:49,122][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:44:49,621][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:44:50,120][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:44:50,619][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:44:51,117][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:44:51,615][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:44:52,114][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:44:52,614][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:44:53,113][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:44:53,612][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:44:54,112][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:44:54,611][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:44:55,110][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:44:55,609][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:44:56,107][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:44:56,606][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:44:57,104][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:44:57,602][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:44:58,101][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:44:58,599][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:44:59,098][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:44:59,597][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:45:00,096][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:45:00,596][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:45:01,095][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:45:01,592][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:45:02,090][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:45:02,586][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:45:03,083][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:45:03,580][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:45:04,078][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:45:04,576][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:45:05,074][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:45:05,574][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:45:06,073][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:45:06,572][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:45:07,070][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:45:07,569][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:45:08,067][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:45:08,566][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:45:09,064][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:45:09,564][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:45:10,067][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:45:10,567][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:45:11,067][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:45:11,566][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:45:12,065][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:45:12,566][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:45:13,064][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:45:13,565][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:45:14,065][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:45:14,565][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:45:15,065][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:45:15,564][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:45:16,065][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:45:16,565][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:45:17,066][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:45:17,696][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:45:18,441][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:45:18,444][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:45:18,445][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:45:19,101][__main__][INFO] - Iteration 530 took 1m 14s (9.29% Gen, 89.82% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 45m 15s. Estimated total time: 61h 55m 52s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 51s, 500 more iterations: 10h 19m 18s. [2026-03-26 07:45:19,104][__main__][INFO] - Starting iteration 530. [2026-03-26 07:45:19,505][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:45:19,505][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:45:26,367][__main__][INFO] - Number of regex retries in iteration 530: 0 [2026-03-26 07:45:26,368][__main__][INFO] - agents played in iteration 530 are Alice, Bob [2026-03-26 07:45:27,244][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:45:27,289][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:45:27,290][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:45:27,290][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:45:27,899][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:45:28,354][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:45:28,858][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:45:29,359][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:45:29,860][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:45:30,360][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:45:30,860][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:45:31,360][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:45:31,859][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:45:32,359][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:45:32,858][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:45:33,357][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:45:33,856][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:45:34,355][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:45:34,854][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:45:35,352][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:45:35,851][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:45:36,353][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:45:36,854][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:45:37,354][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:45:37,854][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:45:38,355][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:45:38,855][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:45:39,356][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:45:39,856][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:45:40,356][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:45:40,856][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:45:41,356][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:45:41,856][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:45:42,356][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:45:42,857][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:45:43,357][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:45:43,856][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:45:44,356][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:45:44,855][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:45:45,354][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:45:45,854][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:45:46,354][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:45:46,854][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:45:47,353][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:45:47,854][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:45:48,356][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:45:48,857][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:45:49,357][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:45:49,858][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:45:50,359][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:45:50,861][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:45:51,363][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:45:51,863][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:45:52,363][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:45:52,862][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:45:53,361][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:45:53,862][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:45:54,362][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:45:54,862][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:45:55,363][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:45:55,863][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:45:56,362][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:45:56,861][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:45:57,362][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:45:57,863][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:45:58,363][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:45:58,862][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:45:59,361][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:45:59,861][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:46:00,361][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:46:00,861][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:46:01,361][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:46:01,862][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:46:02,362][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:46:02,861][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:46:03,361][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:46:03,861][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:46:04,359][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:46:04,858][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:46:05,356][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:46:05,855][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:46:06,354][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:46:06,852][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:46:07,351][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:46:07,849][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:46:08,349][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:46:08,849][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:46:09,349][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:46:09,849][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:46:10,349][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:46:10,849][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:46:11,348][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:46:11,848][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:46:12,348][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:46:12,847][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:46:13,347][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:46:13,846][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:46:14,345][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:46:14,843][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:46:15,343][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:46:15,844][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:46:16,344][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:46:16,844][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:46:17,342][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:46:17,841][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:46:18,342][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:46:18,841][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:46:19,340][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:46:19,840][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:46:20,341][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:46:20,842][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:46:21,342][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:46:21,842][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:46:22,342][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:46:22,843][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:46:23,342][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:46:23,842][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:46:24,341][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:46:24,839][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:46:25,339][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:46:25,838][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:46:26,338][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:46:26,837][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:46:27,336][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:46:27,835][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:46:28,335][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:46:28,834][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:46:29,335][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:46:29,834][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:46:30,335][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:46:30,835][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:46:31,334][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:46:31,834][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 07:46:32,467][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 07:46:33,202][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:46:33,206][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:46:33,208][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:46:34,040][__main__][INFO] - Iteration 531 took 1m 14s (9.21% Gen, 89.67% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 54m 55s. Estimated total time: 62h 6m 47s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 13s, 500 more iterations: 10h 21m 7s. [2026-03-26 07:46:34,042][__main__][INFO] - Starting iteration 531. [2026-03-26 07:46:34,441][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:46:34,441][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:46:36,650][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:46:41,486][__main__][INFO] - Number of regex retries in iteration 531: 1 [2026-03-26 07:46:41,486][__main__][INFO] - agents played in iteration 531 are Alice, Bob [2026-03-26 07:46:42,372][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:46:42,417][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:46:42,418][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:46:42,418][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:46:43,022][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:46:43,477][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:46:43,981][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:46:44,481][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:46:44,980][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:46:45,480][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:46:45,979][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:46:46,478][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:46:46,978][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:46:47,480][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:46:47,983][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:46:48,486][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:46:48,987][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:46:49,488][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:46:49,988][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:46:50,488][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:46:50,989][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:46:51,488][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:46:51,989][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:46:52,490][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:46:52,990][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:46:53,489][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:46:53,988][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:46:54,487][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:46:54,987][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:46:55,486][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:46:55,986][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:46:56,485][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:46:56,985][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:46:57,484][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:46:57,983][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:46:58,482][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:46:58,981][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:46:59,480][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:46:59,979][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:47:00,478][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:47:00,976][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:47:01,475][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:47:01,974][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:47:02,474][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:47:02,974][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:47:03,473][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:47:03,972][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:47:04,470][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:47:04,968][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:47:05,466][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:47:05,965][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:47:06,463][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:47:06,963][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:47:07,463][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:47:07,963][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:47:08,463][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:47:08,963][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:47:09,463][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:47:09,963][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:47:10,464][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:47:10,964][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:47:11,463][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:47:11,962][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:47:12,461][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:47:12,961][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:47:13,461][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:47:13,960][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:47:14,460][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:47:14,960][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:47:15,459][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:47:15,959][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:47:16,459][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:47:16,959][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:47:17,459][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:47:17,959][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:47:18,461][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:47:18,960][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:47:19,461][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:47:19,962][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:47:20,464][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:47:20,964][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:47:21,465][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:47:21,965][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:47:22,465][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:47:22,966][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:47:23,465][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:47:23,966][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:47:24,465][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:47:24,964][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:47:25,463][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:47:25,962][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:47:26,461][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:47:26,959][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:47:27,459][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:47:27,960][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:47:28,459][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:47:28,960][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:47:29,460][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:47:29,961][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:47:30,460][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:47:30,961][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:47:31,460][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:47:31,960][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:47:32,459][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:47:32,958][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:47:33,457][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:47:33,957][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:47:34,457][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:47:34,956][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:47:35,455][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:47:35,954][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:47:36,453][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:47:36,952][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:47:37,451][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:47:37,950][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:47:38,449][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:47:38,947][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:47:39,448][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:47:39,948][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:47:40,448][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:47:40,947][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:47:41,447][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:47:41,948][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:47:42,448][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:47:42,949][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:47:43,448][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:47:43,948][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:47:44,447][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:47:44,946][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:47:45,448][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:47:45,950][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:47:46,451][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:47:46,951][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:47:47,580][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:47:48,334][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:47:48,336][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:47:48,338][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:47:49,205][__main__][INFO] - Iteration 532 took 1m 14s (9.42% Gen, 89.42% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 51h 5m 7s. Estimated total time: 62h 18m 14s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 36s, 500 more iterations: 10h 23m 2s. [2026-03-26 07:47:49,207][__main__][INFO] - Starting iteration 532. [2026-03-26 07:47:49,606][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:47:49,607][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:47:51,702][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:47:55,058][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:47:56,091][__main__][INFO] - Number of regex retries in iteration 532: 2 [2026-03-26 07:47:56,092][__main__][INFO] - agents played in iteration 532 are Alice, Bob [2026-03-26 07:47:56,965][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:47:57,010][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:47:57,010][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:47:57,011][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:47:57,622][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:47:58,078][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:47:58,583][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:47:59,084][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:47:59,584][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:48:00,083][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:48:00,583][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:48:01,083][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:48:01,583][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:48:02,082][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:48:02,581][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:48:03,079][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:48:03,578][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:48:04,077][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:48:04,576][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:48:05,074][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:48:05,573][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:48:06,073][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:48:06,574][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:48:07,076][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:48:07,576][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:48:08,077][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:48:08,576][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:48:09,074][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:48:09,573][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:48:10,071][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:48:10,570][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:48:11,069][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:48:11,568][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:48:12,066][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:48:12,564][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:48:13,062][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:48:13,560][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:48:14,060][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:48:14,560][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:48:15,060][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:48:15,559][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:48:16,059][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:48:16,558][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:48:17,058][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:48:17,557][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:48:18,056][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:48:18,559][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:48:19,058][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:48:19,557][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:48:20,057][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:48:20,556][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:48:21,056][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:48:21,555][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:48:22,059][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:48:22,560][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:48:23,061][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:48:23,562][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:48:24,063][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:48:24,564][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:48:25,065][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:48:25,566][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:48:26,070][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:48:26,572][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:48:27,074][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:48:27,576][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:48:28,079][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:48:28,581][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:48:29,084][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:48:29,586][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:48:30,086][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:48:30,585][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:48:31,086][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:48:31,585][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:48:32,085][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:48:32,585][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:48:33,085][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:48:33,585][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:48:34,083][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:48:34,581][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:48:35,080][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:48:35,579][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:48:36,078][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:48:36,577][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:48:37,075][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:48:37,573][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:48:38,072][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:48:38,571][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:48:39,069][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:48:39,568][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:48:40,067][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:48:40,566][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:48:41,064][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:48:41,563][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:48:42,062][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:48:42,560][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:48:43,059][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:48:43,558][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:48:44,057][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:48:44,556][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:48:45,054][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:48:45,552][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:48:46,052][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:48:46,553][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:48:47,054][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:48:47,553][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:48:48,053][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:48:48,554][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:48:49,054][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:48:49,555][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:48:50,055][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:48:50,554][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:48:51,054][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:48:51,553][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:48:52,051][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:48:52,549][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:48:53,048][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:48:53,546][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:48:54,047][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:48:54,548][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:48:55,049][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:48:55,550][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:48:56,051][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:48:56,553][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:48:57,056][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:48:57,557][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:48:58,059][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:48:58,562][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:48:59,064][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:48:59,566][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:49:00,070][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:49:00,574][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:49:01,076][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:49:01,578][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 07:49:02,224][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.53%, ΔTime: 00:01:04 [2026-03-26 07:49:02,975][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:49:02,978][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:49:02,979][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:49:03,633][__main__][INFO] - Iteration 533 took 1m 14s (8.76% Gen, 90.35% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 27m 1s. Estimated total time: 61h 41m 22s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 22s, 500 more iterations: 10h 16m 53s. [2026-03-26 07:49:03,635][__main__][INFO] - Starting iteration 533. [2026-03-26 07:49:04,036][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:49:04,037][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:49:11,185][__main__][INFO] - Number of regex retries in iteration 533: 0 [2026-03-26 07:49:11,185][__main__][INFO] - agents played in iteration 533 are Alice, Bob [2026-03-26 07:49:12,070][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:49:12,115][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:49:12,116][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:49:12,116][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:49:12,728][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:49:13,178][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:49:13,682][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:49:14,181][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:49:14,678][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:49:15,176][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:49:15,677][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:49:16,177][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:49:16,674][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:49:17,175][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:49:17,674][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:49:18,173][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:49:18,673][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:49:19,172][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:49:19,670][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:49:20,169][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:49:20,668][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:49:21,167][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:49:21,667][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:49:22,166][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:49:22,665][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:49:23,165][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:49:23,664][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:49:24,163][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:49:24,662][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:49:25,163][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:49:25,663][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:49:26,163][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:49:26,662][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:49:27,162][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:49:27,663][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:49:28,162][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:49:28,662][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:49:29,161][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:49:29,662][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:49:30,161][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:49:30,662][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:49:31,161][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:49:31,661][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:49:32,161][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:49:32,661][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:49:33,162][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:49:33,664][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:49:34,165][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:49:34,665][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:49:35,165][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:49:35,665][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:49:36,165][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:49:36,665][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:49:37,164][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:49:37,663][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:49:38,163][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:49:38,662][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:49:39,161][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:49:39,661][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:49:40,160][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:49:40,658][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:49:41,158][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:49:41,658][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:49:42,158][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:49:42,657][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:49:43,158][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:49:43,658][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:49:44,158][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:49:44,658][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:49:45,156][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:49:45,654][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:49:46,151][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:49:46,648][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:49:47,145][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:49:47,643][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:49:48,141][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:49:48,640][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:49:49,140][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:49:49,640][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:49:50,138][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:49:50,637][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:49:51,137][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:49:51,637][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:49:52,136][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:49:52,637][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:49:53,136][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:49:53,636][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:49:54,135][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:49:54,635][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:49:55,136][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:49:55,637][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:49:56,138][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:49:56,637][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:49:57,137][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:49:57,638][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:49:58,138][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:49:58,638][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:49:59,137][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:49:59,637][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:50:00,137][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:50:00,636][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:50:01,136][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:50:01,636][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:50:02,136][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:50:02,636][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:50:03,136][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:50:03,636][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:50:04,135][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:50:04,634][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:50:05,134][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:50:05,634][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:50:06,134][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:50:06,637][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:50:07,138][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:50:07,639][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:50:08,139][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:50:08,640][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:50:09,138][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:50:09,636][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:50:10,136][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:50:10,635][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:50:11,133][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:50:11,631][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:50:12,128][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:50:12,626][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:50:13,125][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:50:13,624][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:50:14,123][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:50:14,622][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:50:15,122][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:50:15,622][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:50:16,121][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:50:16,620][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 07:50:17,260][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.25%, ΔTime: 00:01:04 [2026-03-26 07:50:17,995][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:50:17,998][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:50:17,999][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:50:18,653][__main__][INFO] - Iteration 534 took 1m 14s (9.58% Gen, 89.54% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 55m 17s. Estimated total time: 62h 10m 54s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 21s, 500 more iterations: 10h 21m 49s. [2026-03-26 07:50:18,656][__main__][INFO] - Starting iteration 534. [2026-03-26 07:50:19,056][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:50:19,057][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:50:19,639][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:50:22,886][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:50:23,459][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:50:25,488][__main__][INFO] - Number of regex retries in iteration 534: 3 [2026-03-26 07:50:25,488][__main__][INFO] - agents played in iteration 534 are Alice, Bob [2026-03-26 07:50:26,364][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:50:26,410][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:50:26,410][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:50:26,411][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:50:27,019][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:50:27,474][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:50:27,979][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:50:28,478][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:50:28,977][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:50:29,476][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:50:29,976][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:50:30,475][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:50:30,973][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:50:31,472][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:50:31,971][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:50:32,470][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:50:32,968][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:50:33,467][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:50:33,966][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:50:34,464][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:50:34,962][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:50:35,464][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:50:35,966][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:50:36,467][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:50:36,968][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:50:37,468][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:50:37,969][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:50:38,469][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:50:38,971][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:50:39,470][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:50:39,969][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:50:40,469][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:50:40,969][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:50:41,468][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:50:41,967][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:50:42,466][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:50:42,965][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:50:43,465][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:50:43,964][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:50:44,464][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:50:44,963][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:50:45,465][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:50:45,968][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:50:46,468][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:50:46,967][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:50:47,467][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:50:47,968][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:50:48,471][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:50:48,970][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:50:49,470][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:50:49,970][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:50:50,469][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:50:50,970][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:50:51,470][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:50:51,970][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:50:52,469][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:50:52,970][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:50:53,470][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:50:53,970][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:50:54,470][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:50:54,971][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:50:55,472][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:50:55,972][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:50:56,473][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:50:56,975][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:50:57,477][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:50:57,978][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:50:58,478][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:50:58,978][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:50:59,477][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:50:59,976][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:51:00,476][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:51:00,975][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:51:01,475][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:51:01,972][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:51:02,471][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:51:02,969][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:51:03,469][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:51:03,969][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:51:04,467][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:51:04,967][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:51:05,466][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:51:05,965][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:51:06,464][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:51:06,962][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:51:07,464][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:51:07,965][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:51:08,467][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:51:08,967][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:51:09,470][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:51:09,971][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:51:10,472][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:51:10,973][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:51:11,474][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:51:11,972][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:51:12,470][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:51:12,969][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:51:13,467][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:51:13,966][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:51:14,464][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:51:14,964][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:51:15,464][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:51:15,963][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:51:16,462][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:51:16,961][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:51:17,462][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:51:17,961][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:51:18,464][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:51:18,964][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:51:19,464][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:51:19,963][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:51:20,462][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:51:20,962][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:51:21,462][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:51:21,962][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:51:22,462][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:51:22,962][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:51:23,462][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:51:23,962][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:51:24,462][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:51:24,962][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:51:25,463][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:51:25,964][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:51:26,463][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:51:26,963][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:51:27,463][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:51:27,963][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:51:28,463][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:51:28,964][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:51:29,465][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:51:29,964][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:51:30,464][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:51:30,964][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 07:51:31,597][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:51:32,336][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:51:32,339][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:51:32,340][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:51:32,988][__main__][INFO] - Iteration 535 took 1m 13s (8.70% Gen, 90.42% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 19m 47s. Estimated total time: 61h 36m 38s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 13s, 500 more iterations: 10h 16m 6s. [2026-03-26 07:51:32,990][__main__][INFO] - Starting iteration 535. [2026-03-26 07:51:33,393][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:51:33,394][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:51:38,449][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 0 books, 30 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:51:40,152][__main__][INFO] - Number of regex retries in iteration 535: 1 [2026-03-26 07:51:40,153][__main__][INFO] - agents played in iteration 535 are Alice, Bob [2026-03-26 07:51:41,033][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:51:41,078][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:51:41,079][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:51:41,079][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:51:41,685][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:51:42,139][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:51:42,643][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:51:43,142][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:51:43,641][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:51:44,140][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:51:44,639][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:51:45,137][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:51:45,636][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:51:46,136][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:51:46,634][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:51:47,132][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:51:47,630][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:51:48,129][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:51:48,630][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:51:49,129][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:51:49,628][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:51:50,128][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:51:50,627][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:51:51,126][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:51:51,625][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:51:52,124][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:51:52,624][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:51:53,124][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:51:53,623][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:51:54,122][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:51:54,621][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:51:55,121][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:51:55,620][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:51:56,119][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:51:56,618][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:51:57,117][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:51:57,615][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:51:58,113][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:51:58,612][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:51:59,110][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:51:59,607][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:52:00,105][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:52:00,604][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:52:01,102][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:52:01,600][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:52:02,099][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:52:02,598][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:52:03,098][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:52:03,598][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:52:04,098][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:52:04,598][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:52:05,099][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:52:05,598][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:52:06,096][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:52:06,593][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:52:07,091][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:52:07,589][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:52:08,087][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:52:08,585][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:52:09,084][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:52:09,583][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:52:10,082][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:52:10,582][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:52:11,081][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:52:11,581][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:52:12,079][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:52:12,579][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:52:13,079][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:52:13,578][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:52:14,075][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:52:14,573][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:52:15,070][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:52:15,571][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:52:16,071][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:52:16,570][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:52:17,068][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:52:17,566][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:52:18,066][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:52:18,566][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:52:19,065][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:52:19,564][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:52:20,063][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:52:20,563][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:52:21,063][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:52:21,562][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:52:22,062][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:52:22,561][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:52:23,062][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:52:23,562][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:52:24,062][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:52:24,562][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:52:25,063][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:52:25,562][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:52:26,062][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:52:26,562][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:52:27,062][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:52:27,561][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:52:28,062][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:52:28,561][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:52:29,061][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:52:29,560][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:52:30,059][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:52:30,557][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:52:31,056][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:52:31,555][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:52:32,053][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:52:32,551][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:52:33,048][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:52:33,545][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:52:34,044][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:52:34,544][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:52:35,042][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:52:35,543][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:52:36,042][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:52:36,543][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:52:37,043][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:52:37,543][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:52:38,041][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:52:38,540][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:52:39,039][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:52:39,537][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:52:40,035][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:52:40,532][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:52:41,030][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:52:41,527][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:52:42,025][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:52:42,523][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:52:43,022][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:52:43,521][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:52:44,019][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:52:44,517][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:52:45,016][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:52:45,514][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:52:46,149][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.20%, ΔTime: 00:01:04 [2026-03-26 07:52:46,890][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:52:46,892][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:52:46,894][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:52:47,611][__main__][INFO] - Iteration 536 took 1m 14s (9.11% Gen, 89.92% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 32m 50s. Estimated total time: 61h 50m 56s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 29s. [2026-03-26 07:52:47,614][__main__][INFO] - Starting iteration 536. [2026-03-26 07:52:48,018][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:52:48,018][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:52:51,501][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:52:54,683][__main__][INFO] - Number of regex retries in iteration 536: 1 [2026-03-26 07:52:54,684][__main__][INFO] - agents played in iteration 536 are Alice, Bob [2026-03-26 07:52:55,564][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:52:55,609][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:52:55,610][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:52:55,610][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:52:56,219][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:52:56,672][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:52:57,177][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:52:57,676][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:52:58,176][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:52:58,675][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:52:59,174][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:52:59,675][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:53:00,176][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:53:00,678][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:53:01,180][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:53:01,681][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:53:02,182][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:53:02,683][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:53:03,184][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:53:03,685][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:53:04,185][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:53:04,685][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:53:05,185][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:53:05,687][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:53:06,186][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:53:06,686][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:53:07,186][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:53:07,686][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:53:08,186][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:53:08,686][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:53:09,185][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:53:09,684][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:53:10,184][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:53:10,683][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:53:11,183][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:53:11,683][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:53:12,183][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:53:12,685][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:53:13,185][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:53:13,685][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:53:14,185][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:53:14,686][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:53:15,186][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:53:15,686][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:53:16,186][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:53:16,685][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:53:17,185][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:53:17,685][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:53:18,186][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:53:18,686][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:53:19,186][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:53:19,685][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:53:20,185][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:53:20,684][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:53:21,182][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:53:21,681][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:53:22,179][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:53:22,678][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:53:23,176][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:53:23,673][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:53:24,170][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:53:24,669][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:53:25,169][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:53:25,668][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:53:26,167][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:53:26,666][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:53:27,165][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:53:27,663][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:53:28,163][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:53:28,662][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:53:29,162][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:53:29,661][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:53:30,160][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:53:30,660][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:53:31,160][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:53:31,658][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:53:32,158][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:53:32,659][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:53:33,161][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:53:33,661][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:53:34,161][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:53:34,662][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:53:35,164][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:53:35,665][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:53:36,166][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:53:36,665][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:53:37,164][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:53:37,663][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:53:38,162][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:53:38,662][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:53:39,162][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:53:39,662][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:53:40,161][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:53:40,660][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:53:41,160][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:53:41,660][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:53:42,160][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:53:42,659][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:53:43,158][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:53:43,658][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:53:44,157][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:53:44,658][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:53:45,159][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:53:45,663][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:53:46,165][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:53:46,665][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:53:47,165][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:53:47,666][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:53:48,167][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:53:48,668][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:53:49,168][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:53:49,668][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:53:50,168][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:53:50,668][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:53:51,168][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:53:51,667][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:53:52,167][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:53:52,666][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:53:53,166][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:53:53,665][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:53:54,164][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:53:54,663][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:53:55,163][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:53:55,662][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:53:56,161][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:53:56,660][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:53:57,158][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:53:57,657][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:53:58,156][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:53:58,654][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:53:59,151][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:53:59,649][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:54:00,146][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 07:54:00,779][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 07:54:01,804][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:54:01,806][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:54:01,808][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:54:02,464][__main__][INFO] - Iteration 537 took 1m 14s (8.95% Gen, 90.16% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 50h 43m 0s. Estimated total time: 62h 2m 21s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 4s, 500 more iterations: 10h 20m 23s. [2026-03-26 07:54:02,467][__main__][INFO] - Starting iteration 537. [2026-03-26 07:54:02,866][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:54:02,867][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:54:03,459][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:54:09,719][__main__][INFO] - Number of regex retries in iteration 537: 1 [2026-03-26 07:54:09,720][__main__][INFO] - agents played in iteration 537 are Alice, Bob [2026-03-26 07:54:10,598][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:54:10,643][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:54:10,644][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:54:10,644][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:54:11,268][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:54:11,722][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:54:12,226][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:54:12,725][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:54:13,224][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:54:13,723][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:54:14,222][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:54:14,722][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:54:15,221][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:54:15,723][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:54:16,224][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:54:16,723][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:54:17,223][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:54:17,724][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:54:18,227][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:54:18,728][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:54:19,227][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:54:19,729][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:54:20,229][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:54:20,729][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:54:21,229][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:54:21,730][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:54:22,230][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:54:22,730][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:54:23,230][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:54:23,730][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:54:24,229][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:54:24,729][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:54:25,229][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:54:25,729][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:54:26,229][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:54:26,729][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:54:27,228][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:54:27,727][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:54:28,227][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:54:28,727][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:54:29,227][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:54:29,727][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:54:30,227][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:54:30,726][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:54:31,226][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:54:31,726][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:54:32,226][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:54:32,726][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:54:33,226][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:54:33,727][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:54:34,227][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:54:34,726][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:54:35,226][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:54:35,746][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:54:36,251][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:54:36,752][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:54:37,252][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:54:37,753][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:54:38,254][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:54:38,755][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:54:39,256][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:54:39,756][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:54:40,255][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:54:40,757][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:54:41,258][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:54:41,758][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:54:42,258][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:54:42,757][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:54:43,257][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:54:43,758][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:54:44,258][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:54:44,757][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:54:45,256][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:54:45,756][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:54:46,255][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:54:46,754][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:54:47,254][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:54:47,757][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:54:48,260][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:54:48,760][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:54:49,259][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:54:49,759][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:54:50,260][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:54:50,760][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:54:51,260][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:54:51,761][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:54:52,263][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:54:52,763][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:54:53,264][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:54:53,765][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:54:54,266][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:54:54,767][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:54:55,269][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:54:55,769][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:54:56,269][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:54:56,768][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:54:57,268][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:54:57,769][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:54:58,269][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:54:58,769][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:54:59,269][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:54:59,769][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:55:00,269][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:55:00,769][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:55:01,269][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:55:01,769][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:55:02,270][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:55:02,769][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:55:03,270][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:55:03,770][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:55:04,270][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:55:04,771][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:55:05,271][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:55:05,773][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:55:06,273][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:55:06,773][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:55:07,273][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:55:07,776][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:55:08,278][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:55:08,780][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:55:09,280][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:55:09,781][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:55:10,283][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:55:10,784][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:55:11,284][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:55:11,784][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:55:12,283][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:55:12,783][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:55:13,282][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:55:13,782][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:55:14,282][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:55:14,782][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:55:15,283][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 07:55:15,924][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.22%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 07:55:16,655][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:55:16,658][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:55:16,659][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:55:17,309][__main__][INFO] - Iteration 538 took 1m 14s (9.21% Gen, 89.92% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 41m 36s. Estimated total time: 62h 2m 11s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 4s, 500 more iterations: 10h 20m 21s. [2026-03-26 07:55:17,312][__main__][INFO] - Starting iteration 538. [2026-03-26 07:55:17,712][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:55:17,713][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:55:18,293][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:55:24,168][__main__][INFO] - Number of regex retries in iteration 538: 1 [2026-03-26 07:55:24,169][__main__][INFO] - agents played in iteration 538 are Alice, Bob [2026-03-26 07:55:25,047][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:55:25,092][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:55:25,093][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:55:25,094][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:55:25,700][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:55:26,157][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:55:26,661][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:55:27,161][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:55:27,661][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:55:28,161][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:55:28,661][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:55:29,160][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:55:29,661][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:55:30,163][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:55:30,665][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:55:31,167][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:55:31,668][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:55:32,169][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:55:32,669][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:55:33,169][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:55:33,670][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:55:34,169][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:55:34,668][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:55:35,167][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:55:35,666][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:55:36,165][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:55:36,664][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:55:37,163][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:55:37,664][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:55:38,164][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:55:38,665][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:55:39,165][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:55:39,665][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:55:40,165][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:55:40,665][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:55:41,166][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:55:41,667][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:55:42,166][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:55:42,664][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:55:43,163][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:55:43,662][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:55:44,160][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:55:44,659][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:55:45,158][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:55:45,657][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:55:46,158][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:55:46,658][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:55:47,158][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:55:47,659][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:55:48,162][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:55:48,663][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:55:49,163][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:55:49,664][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:55:50,165][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:55:50,665][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:55:51,166][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:55:51,667][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:55:52,167][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:55:52,667][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:55:53,167][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:55:53,668][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:55:54,167][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:55:54,665][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:55:55,165][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:55:55,665][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:55:56,164][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:55:56,662][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:55:57,161][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:55:57,660][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:55:58,160][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:55:58,660][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:55:59,160][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:55:59,659][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:56:00,159][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:56:00,657][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:56:01,157][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:56:01,656][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:56:02,156][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:56:02,659][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:56:03,161][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:56:03,662][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:56:04,162][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:56:04,663][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:56:05,165][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:56:05,666][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:56:06,166][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:56:06,666][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:56:07,165][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:56:07,664][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:56:08,163][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:56:08,663][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:56:09,163][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:56:09,662][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:56:10,162][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:56:10,662][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:56:11,162][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:56:11,662][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:56:12,163][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:56:12,664][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:56:13,164][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:56:13,665][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:56:14,164][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:56:14,663][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:56:15,162][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:56:15,661][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:56:16,160][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:56:16,659][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:56:17,159][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:56:17,658][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:56:18,159][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:56:18,660][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:56:19,159][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:56:19,659][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:56:20,159][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:56:20,659][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:56:21,159][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:56:21,659][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:56:22,159][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:56:22,660][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:56:23,161][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:56:23,661][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:56:24,161][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:56:24,661][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:56:25,163][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:56:25,664][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:56:26,163][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:56:26,663][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:56:27,162][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:56:27,662][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:56:28,161][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:56:28,659][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:56:29,158][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:56:29,658][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 07:56:30,289][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 07:56:31,026][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:56:31,028][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:56:31,030][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:56:31,677][__main__][INFO] - Iteration 539 took 1m 13s (8.73% Gen, 90.39% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 16m 28s. Estimated total time: 61h 38m 17s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 16s, 500 more iterations: 10h 16m 22s. [2026-03-26 07:56:31,680][__main__][INFO] - Starting iteration 539. [2026-03-26 07:56:32,080][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:56:32,080][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:56:32,668][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:56:38,194][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 07:56:39,249][__main__][INFO] - Number of regex retries in iteration 539: 2 [2026-03-26 07:56:39,249][__main__][INFO] - agents played in iteration 539 are Alice, Bob [2026-03-26 07:56:40,137][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:56:40,182][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:56:40,183][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:56:40,183][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:56:40,789][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:56:41,245][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:56:41,749][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:56:42,248][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:56:42,746][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:56:43,245][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:56:43,745][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:56:44,245][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:56:44,744][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:56:45,244][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:56:45,746][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:56:46,247][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:56:46,746][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:56:47,246][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:56:47,746][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:56:48,249][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:56:48,749][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:56:49,248][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:56:49,746][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:56:50,244][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:56:50,743][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:56:51,243][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:56:51,742][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:56:52,241][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:56:52,739][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:56:53,240][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:56:53,740][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:56:54,240][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:56:54,741][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:56:55,242][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:56:55,743][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:56:56,243][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:56:56,743][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:56:57,242][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:56:57,742][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:56:58,242][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:56:58,742][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:56:59,241][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:56:59,740][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:57:00,241][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:57:00,741][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:57:01,240][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:57:01,740][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:57:02,240][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:57:02,739][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:57:03,238][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:57:03,737][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:57:04,236][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:57:04,738][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:57:05,239][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:57:05,739][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:57:06,239][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:57:06,739][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:57:07,239][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:57:07,739][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:57:08,238][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:57:08,740][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:57:09,242][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:57:09,743][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:57:10,244][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:57:10,745][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:57:11,246][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:57:11,747][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:57:12,248][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:57:12,750][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:57:13,251][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:57:13,752][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:57:14,252][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:57:14,752][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:57:15,252][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:57:15,751][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:57:16,251][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:57:16,749][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:57:17,249][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:57:17,747][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:57:18,249][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:57:18,749][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:57:19,248][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:57:19,748][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:57:20,247][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:57:20,747][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:57:21,246][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:57:21,745][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:57:22,245][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:57:22,745][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:57:23,246][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:57:23,745][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:57:24,245][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:57:24,744][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:57:25,245][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:57:25,745][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:57:26,247][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:57:26,748][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:57:27,247][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:57:27,748][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:57:28,248][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:57:28,748][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:57:29,247][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:57:29,747][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:57:30,248][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:57:30,747][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:57:31,247][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:57:31,746][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:57:32,246][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:57:32,747][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:57:33,246][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:57:33,745][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:57:34,245][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:57:34,743][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:57:35,243][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:57:35,742][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:57:36,241][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:57:36,740][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:57:37,240][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:57:37,740][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:57:38,240][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:57:38,739][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:57:39,239][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:57:39,739][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:57:40,239][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:57:40,739][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:57:41,239][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:57:41,739][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:57:42,240][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:57:42,742][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:57:43,243][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:57:43,744][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:57:44,246][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:57:44,747][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21749 tokens. [2026-03-26 07:57:45,381][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.32%, Current % of VRAM taken: 60.68%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 07:57:46,111][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:57:46,114][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:57:46,115][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:57:46,782][__main__][INFO] - Iteration 540 took 1m 14s (9.60% Gen, 89.51% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 52m 4s. Estimated total time: 62h 15m 9s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 30s, 500 more iterations: 10h 22m 31s. [2026-03-26 07:57:46,784][__main__][INFO] - Starting iteration 540. [2026-03-26 07:57:47,185][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:57:47,185][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:57:53,476][__main__][INFO] - Number of regex retries in iteration 540: 0 [2026-03-26 07:57:53,477][__main__][INFO] - agents played in iteration 540 are Alice, Bob [2026-03-26 07:57:54,353][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:57:54,398][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.53%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:57:54,399][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:57:54,399][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:57:55,011][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:57:55,465][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:57:55,968][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:57:56,467][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:57:56,967][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:57:57,467][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:57:57,967][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:57:58,466][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:57:58,965][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:57:59,464][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:57:59,964][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:58:00,463][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:58:00,963][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:58:01,462][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:58:01,961][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:58:02,459][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:58:02,958][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:58:03,460][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:58:03,961][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:58:04,462][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:58:04,962][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:58:05,462][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:58:05,962][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:58:06,462][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:58:06,962][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:58:07,461][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:58:07,960][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:58:08,460][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:58:08,959][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:58:09,458][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:58:09,957][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:58:10,457][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:58:10,957][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:58:11,460][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:58:11,963][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:58:12,465][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:58:12,966][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:58:13,466][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:58:13,967][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:58:14,468][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:58:14,968][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:58:15,470][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:58:15,972][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:58:16,472][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:58:16,970][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:58:17,469][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:58:17,968][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:58:18,470][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:58:18,969][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:58:19,467][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:58:19,966][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:58:20,465][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:58:20,964][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:58:21,465][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:58:21,964][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:58:22,463][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:58:22,962][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:58:23,462][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:58:23,963][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:58:24,464][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:58:24,965][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:58:25,466][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:58:25,967][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:58:26,467][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:58:26,968][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:58:27,467][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:58:27,966][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:58:28,466][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:58:28,966][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:58:29,465][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:58:29,964][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:58:30,465][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:58:30,965][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:58:31,463][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:58:31,962][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:58:32,461][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:58:32,960][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:58:33,459][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:58:33,959][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:58:34,458][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:58:34,956][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:58:35,456][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:58:35,956][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:58:36,457][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:58:36,957][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:58:37,456][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:58:37,955][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:58:38,455][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:58:38,954][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:58:39,453][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:58:39,952][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:58:40,450][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:58:40,950][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:58:41,448][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:58:41,946][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:58:42,444][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:58:42,943][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:58:43,443][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:58:43,943][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:58:44,444][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:58:44,944][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 07:58:45,443][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 07:58:45,943][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 07:58:46,444][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 07:58:46,945][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 07:58:47,444][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 07:58:47,943][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 07:58:48,444][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 07:58:48,942][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 07:58:49,440][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 07:58:49,939][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 07:58:50,438][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 07:58:50,938][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 07:58:51,437][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 07:58:51,936][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 07:58:52,434][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 07:58:52,931][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 07:58:53,429][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 07:58:53,926][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 07:58:54,423][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 07:58:54,921][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 07:58:55,422][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 07:58:55,922][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 07:58:56,422][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 07:58:56,921][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 07:58:57,421][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 07:58:57,920][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 07:58:58,420][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 07:58:58,920][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 07:58:59,558][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 07:59:00,291][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 07:59:00,293][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 07:59:00,295][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 07:59:01,120][__main__][INFO] - Iteration 541 took 1m 13s (8.51% Gen, 90.37% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 12m 28s. Estimated total time: 61h 36m 47s. Time estimates for 10 more iterations: 12m 19s, 100 more iterations: 2h 3m 13s, 500 more iterations: 10h 16m 7s. [2026-03-26 07:59:01,122][__main__][INFO] - Starting iteration 541. [2026-03-26 07:59:01,523][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 07:59:01,523][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 07:59:08,158][__main__][INFO] - Number of regex retries in iteration 541: 0 [2026-03-26 07:59:08,159][__main__][INFO] - agents played in iteration 541 are Alice, Bob [2026-03-26 07:59:09,048][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:59:09,093][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 07:59:09,093][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 07:59:09,094][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 07:59:09,711][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 07:59:10,167][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 07:59:10,671][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 07:59:11,171][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 07:59:11,671][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 07:59:12,171][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 07:59:12,671][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 07:59:13,171][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 07:59:13,671][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 07:59:14,172][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 07:59:14,673][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 07:59:15,174][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 07:59:15,674][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 07:59:16,174][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 07:59:16,673][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 07:59:17,174][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 07:59:17,673][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 07:59:18,172][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 07:59:18,671][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 07:59:19,170][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 07:59:19,668][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 07:59:20,166][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 07:59:20,665][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 07:59:21,163][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 07:59:21,662][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 07:59:22,161][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 07:59:22,660][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 07:59:23,160][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 07:59:23,660][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 07:59:24,158][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 07:59:24,658][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 07:59:25,158][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 07:59:25,661][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 07:59:26,160][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 07:59:26,659][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 07:59:27,158][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 07:59:27,655][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 07:59:28,153][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 07:59:28,651][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 07:59:29,149][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 07:59:29,647][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 07:59:30,148][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 07:59:30,649][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 07:59:31,148][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 07:59:31,648][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 07:59:32,147][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 07:59:32,646][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 07:59:33,146][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 07:59:33,646][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 07:59:34,145][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 07:59:34,644][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 07:59:35,143][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 07:59:35,642][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 07:59:36,142][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 07:59:36,641][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 07:59:37,142][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 07:59:37,642][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 07:59:38,141][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 07:59:38,638][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 07:59:39,136][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 07:59:39,633][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 07:59:40,131][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 07:59:40,628][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 07:59:41,127][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 07:59:41,626][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 07:59:42,125][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 07:59:42,625][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 07:59:43,125][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 07:59:43,624][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 07:59:44,124][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 07:59:44,623][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 07:59:45,123][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 07:59:45,625][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 07:59:46,127][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 07:59:46,627][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 07:59:47,127][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 07:59:47,627][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 07:59:48,128][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 07:59:48,629][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 07:59:49,130][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 07:59:49,629][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 07:59:50,126][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 07:59:50,624][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 07:59:51,121][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 07:59:51,619][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 07:59:52,119][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 07:59:52,618][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 07:59:53,116][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 07:59:53,613][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 07:59:54,112][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 07:59:54,610][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 07:59:55,110][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 07:59:55,609][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 07:59:56,108][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 07:59:56,607][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 07:59:57,106][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 07:59:57,605][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 07:59:58,103][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 07:59:58,600][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 07:59:59,098][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 07:59:59,594][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:00:00,092][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:00:00,590][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:00:01,088][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:00:01,586][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:00:02,085][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:00:02,586][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:00:03,085][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:00:03,586][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:00:04,086][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:00:04,586][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:00:05,086][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:00:05,586][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:00:06,085][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:00:06,584][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:00:07,082][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:00:07,580][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:00:08,079][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:00:08,579][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:00:09,078][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:00:09,577][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:00:10,077][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:00:10,577][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:00:11,076][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:00:11,574][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:00:12,074][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:00:12,572][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:00:13,070][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:00:13,569][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 08:00:14,200][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.56%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 08:00:14,968][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:00:14,971][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:00:14,972][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:00:15,631][__main__][INFO] - Iteration 542 took 1m 14s (8.95% Gen, 90.15% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 19m 53s. Estimated total time: 61h 45m 26s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 30s, 500 more iterations: 10h 17m 34s. [2026-03-26 08:00:15,633][__main__][INFO] - Starting iteration 542. [2026-03-26 08:00:16,038][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:00:16,039][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:00:19,745][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:00:22,263][__main__][INFO] - Number of regex retries in iteration 542: 1 [2026-03-26 08:00:22,264][__main__][INFO] - agents played in iteration 542 are Alice, Bob [2026-03-26 08:00:23,157][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:00:23,202][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:00:23,203][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:00:23,204][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:00:23,808][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:00:24,264][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:00:24,767][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:00:25,267][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:00:25,767][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:00:26,266][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:00:26,766][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:00:27,266][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:00:27,766][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:00:28,266][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:00:28,764][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:00:29,263][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:00:29,763][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:00:30,261][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:00:30,761][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:00:31,260][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:00:31,759][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:00:32,258][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:00:32,756][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:00:33,254][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:00:33,751][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:00:34,249][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:00:34,747][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:00:35,246][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:00:35,744][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:00:36,244][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:00:36,744][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:00:37,244][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:00:37,743][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:00:38,243][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:00:38,742][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:00:39,242][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:00:39,742][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:00:40,241][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:00:40,739][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:00:41,238][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:00:41,737][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:00:42,237][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:00:42,737][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:00:43,236][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:00:43,735][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:00:44,233][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:00:44,730][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:00:45,228][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:00:45,725][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:00:46,224][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:00:46,722][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:00:47,220][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:00:47,719][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:00:48,217][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:00:48,715][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:00:49,212][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:00:49,708][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:00:50,206][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:00:50,702][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:00:51,200][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:00:51,697][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:00:52,199][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:00:52,699][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:00:53,200][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:00:53,700][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:00:54,200][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:00:54,701][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:00:55,203][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:00:55,703][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:00:56,202][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:00:56,701][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:00:57,201][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:00:57,699][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:00:58,201][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:00:58,700][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:00:59,201][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:00:59,700][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:01:00,198][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:01:00,696][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:01:01,194][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:01:01,692][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:01:02,190][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:01:02,689][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:01:03,188][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:01:03,686][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:01:04,185][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:01:04,682][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:01:05,180][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:01:05,678][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:01:06,177][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:01:06,675][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:01:07,173][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:01:07,672][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:01:08,172][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:01:08,673][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:01:09,172][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:01:09,672][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:01:10,171][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:01:10,670][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:01:11,169][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:01:11,669][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:01:12,168][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:01:12,667][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:01:13,167][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:01:13,666][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:01:14,166][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:01:14,666][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:01:15,166][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:01:15,669][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:01:16,169][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:01:16,667][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:01:17,166][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:01:17,665][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:01:18,165][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:01:18,665][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:01:19,164][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:01:19,663][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:01:20,160][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:01:20,656][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:01:21,153][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:01:21,650][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:01:22,147][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:01:22,644][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:01:23,141][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:01:23,638][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:01:24,139][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:01:24,638][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:01:25,139][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:01:25,641][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:01:26,141][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:01:26,641][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:01:27,140][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:01:27,641][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 08:01:28,271][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 08:01:29,076][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:01:29,078][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:01:29,080][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:01:29,903][__main__][INFO] - Iteration 543 took 1m 13s (8.43% Gen, 90.46% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 6m 29s. Estimated total time: 61h 33m 17s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 6s, 500 more iterations: 10h 15m 32s. [2026-03-26 08:01:29,906][__main__][INFO] - Starting iteration 543. [2026-03-26 08:01:30,306][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:01:30,307][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:01:30,884][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:01:37,417][__main__][INFO] - Number of regex retries in iteration 543: 1 [2026-03-26 08:01:37,418][__main__][INFO] - agents played in iteration 543 are Alice, Bob [2026-03-26 08:01:38,295][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:01:38,339][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:01:38,340][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:01:38,341][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:01:38,959][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:01:39,418][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:01:39,922][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:01:40,423][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:01:40,923][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:01:41,423][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:01:41,923][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:01:42,423][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:01:42,923][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:01:43,424][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:01:43,925][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:01:44,424][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:01:44,924][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:01:45,426][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:01:45,926][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:01:46,426][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:01:46,927][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:01:47,429][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:01:47,930][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:01:48,432][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:01:48,932][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:01:49,432][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:01:49,932][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:01:50,431][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:01:50,932][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:01:51,429][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:01:51,926][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:01:52,423][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:01:52,920][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:01:53,416][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:01:53,913][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:01:54,410][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:01:54,907][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:01:55,409][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:01:55,909][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:01:56,408][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:01:56,908][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:01:57,408][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:01:57,907][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:01:58,407][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:01:58,908][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:01:59,408][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:01:59,907][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:02:00,406][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:02:00,905][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:02:01,406][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:02:01,905][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:02:02,405][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:02:02,904][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:02:03,401][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:02:03,898][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:02:04,395][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:02:04,891][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:02:05,389][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:02:05,886][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:02:06,383][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:02:06,881][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:02:07,381][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:02:07,881][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:02:08,380][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:02:08,879][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:02:09,378][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:02:09,880][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:02:10,380][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:02:10,880][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:02:11,380][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:02:11,880][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:02:12,379][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:02:12,877][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:02:13,377][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:02:13,878][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:02:14,378][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:02:14,877][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:02:15,377][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:02:15,877][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:02:16,376][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:02:16,876][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:02:17,376][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:02:17,877][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:02:18,379][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:02:18,880][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:02:19,380][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:02:19,881][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:02:20,382][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:02:20,882][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:02:21,383][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:02:21,884][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:02:22,383][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:02:22,883][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:02:23,379][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:02:23,875][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:02:24,372][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:02:24,870][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:02:25,369][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:02:25,866][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:02:26,363][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:02:26,860][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:02:27,362][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:02:27,862][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:02:28,362][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:02:28,863][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:02:29,363][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:02:29,864][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:02:30,364][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:02:30,864][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:02:31,364][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:02:31,865][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:02:32,365][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:02:32,865][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:02:33,364][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:02:33,865][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:02:34,365][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:02:34,864][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:02:35,364][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:02:35,864][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:02:36,361][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:02:36,858][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:02:37,355][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:02:37,852][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:02:38,349][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:02:38,846][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:02:39,347][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:02:39,847][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:02:40,347][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:02:40,846][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:02:41,346][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:02:41,845][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:02:42,345][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:02:42,844][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21742 tokens. [2026-03-26 08:02:43,485][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.36%, ΔTime: 00:01:04 [2026-03-26 08:02:44,260][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:02:44,263][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:02:44,264][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:02:44,967][__main__][INFO] - Iteration 544 took 1m 14s (9.52% Gen, 89.53% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 45m 2s. Estimated total time: 62h 13m 5s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 26s, 500 more iterations: 10h 22m 10s. [2026-03-26 08:02:44,970][__main__][INFO] - Starting iteration 544. [2026-03-26 08:02:45,369][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:02:45,370][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:02:52,501][__main__][INFO] - Number of regex retries in iteration 544: 0 [2026-03-26 08:02:52,502][__main__][INFO] - agents played in iteration 544 are Alice, Bob [2026-03-26 08:02:53,382][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:02:53,427][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:02:53,427][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:02:53,428][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:02:54,033][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:02:54,487][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:02:54,990][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:02:55,490][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:02:55,989][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:02:56,488][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:02:56,986][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:02:57,485][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:02:57,984][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:02:58,485][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:02:58,985][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:02:59,486][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:02:59,986][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:03:00,486][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:03:00,987][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:03:01,487][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:03:01,988][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:03:02,487][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:03:02,986][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:03:03,485][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:03:03,984][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:03:04,483][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:03:04,982][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:03:05,481][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:03:05,980][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:03:06,480][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:03:06,981][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:03:07,482][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:03:07,982][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:03:08,481][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:03:08,981][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:03:09,480][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:03:09,980][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:03:10,479][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:03:10,980][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:03:11,482][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:03:11,983][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:03:12,483][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:03:12,983][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:03:13,483][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:03:13,984][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:03:14,483][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:03:14,981][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:03:15,480][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:03:15,979][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:03:16,478][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:03:16,976][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:03:17,474][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:03:17,972][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:03:18,473][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:03:18,971][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:03:19,469][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:03:19,969][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:03:20,469][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:03:20,967][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:03:21,466][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:03:21,965][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:03:22,465][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:03:22,964][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:03:23,463][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:03:23,962][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:03:24,461][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:03:24,960][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:03:25,461][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:03:25,962][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:03:26,461][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:03:26,961][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:03:27,460][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:03:27,960][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:03:28,459][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:03:28,958][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:03:29,456][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:03:29,954][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:03:30,454][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:03:30,953][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:03:31,454][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:03:31,954][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:03:32,454][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:03:32,953][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:03:33,452][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:03:33,953][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:03:34,451][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:03:34,949][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:03:35,447][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:03:35,946][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:03:36,445][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:03:36,943][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:03:37,441][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:03:37,940][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:03:38,441][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:03:38,940][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:03:39,439][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:03:39,939][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:03:40,438][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:03:40,937][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:03:41,435][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:03:41,936][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:03:42,435][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:03:42,935][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:03:43,435][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:03:43,935][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:03:44,435][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:03:44,935][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:03:45,433][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:03:45,932][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:03:46,432][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:03:46,930][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:03:47,427][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:03:47,927][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:03:48,427][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:03:48,926][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:03:49,425][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:03:49,924][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:03:50,423][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:03:50,921][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:03:51,419][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:03:51,917][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:03:52,415][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:03:52,913][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:03:53,411][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:03:53,909][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:03:54,409][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:03:54,908][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:03:55,407][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:03:55,907][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:03:56,406][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:03:56,906][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:03:57,405][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:03:57,905][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 08:03:58,535][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 08:03:59,306][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:03:59,308][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:03:59,310][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:04:00,015][__main__][INFO] - Iteration 545 took 1m 14s (9.56% Gen, 89.50% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 43m 3s. Estimated total time: 62h 12m 20s. Time estimates for 10 more iterations: 12m 26s, 100 more iterations: 2h 4m 24s, 500 more iterations: 10h 22m 3s. [2026-03-26 08:04:00,018][__main__][INFO] - Starting iteration 545. [2026-03-26 08:04:00,417][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:04:00,418][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:04:01,490][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:04:02,163][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:04:06,972][__main__][INFO] - Number of regex retries in iteration 545: 2 [2026-03-26 08:04:06,973][__main__][INFO] - agents played in iteration 545 are Alice, Bob [2026-03-26 08:04:07,852][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:04:07,897][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:04:07,898][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:04:07,898][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:04:08,508][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:04:08,965][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:04:09,470][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:04:09,971][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:04:10,470][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:04:10,970][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:04:11,471][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:04:11,971][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:04:12,475][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:04:12,974][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:04:13,473][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:04:13,972][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:04:14,471][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:04:14,971][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:04:15,473][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:04:15,976][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:04:16,476][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:04:16,976][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:04:17,475][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:04:17,976][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:04:18,479][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:04:18,980][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:04:19,480][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:04:19,979][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:04:20,479][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:04:20,978][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:04:21,479][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:04:21,979][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:04:22,479][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:04:22,979][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:04:23,478][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:04:23,976][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:04:24,475][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:04:24,975][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:04:25,474][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:04:25,973][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:04:26,473][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:04:26,972][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:04:27,471][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:04:27,969][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:04:28,468][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:04:28,968][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:04:29,468][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:04:29,967][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:04:30,467][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:04:30,967][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:04:31,467][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:04:31,967][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:04:32,467][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:04:32,968][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:04:33,467][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:04:33,967][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:04:34,466][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:04:34,966][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:04:35,466][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:04:35,965][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:04:36,465][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:04:36,965][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:04:37,465][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:04:37,964][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:04:38,465][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:04:38,965][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:04:39,465][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:04:39,964][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:04:40,464][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:04:40,964][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:04:41,464][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:04:41,964][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:04:42,465][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:04:42,964][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:04:43,464][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:04:43,963][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:04:44,463][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:04:44,963][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:04:45,462][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:04:45,961][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:04:46,463][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:04:46,962][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:04:47,462][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:04:47,961][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:04:48,463][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:04:48,963][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:04:49,464][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:04:49,964][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:04:50,464][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:04:50,964][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:04:51,464][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:04:51,964][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:04:52,465][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:04:52,965][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:04:53,464][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:04:53,964][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:04:54,464][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:04:54,963][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:04:55,463][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:04:55,962][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:04:56,462][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:04:56,962][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:04:57,463][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:04:57,962][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:04:58,462][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:04:58,962][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:04:59,461][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:04:59,960][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:05:00,461][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:05:00,961][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:05:01,460][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:05:01,959][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:05:02,459][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:05:02,958][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:05:03,456][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:05:03,955][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:05:04,453][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:05:04,953][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:05:05,452][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:05:05,952][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:05:06,451][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:05:06,950][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:05:07,449][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:05:07,949][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:05:08,448][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:05:08,948][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:05:09,448][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:05:09,947][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:05:10,447][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:05:10,946][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:05:11,446][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:05:11,945][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:05:12,445][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:05:13,074][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 08:05:13,811][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:05:13,813][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:05:13,815][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:05:14,476][__main__][INFO] - Iteration 546 took 1m 14s (8.85% Gen, 90.25% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 12m 25s. Estimated total time: 61h 42m 58s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 25s, 500 more iterations: 10h 17m 9s. [2026-03-26 08:05:14,478][__main__][INFO] - Starting iteration 546. [2026-03-26 08:05:14,876][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:05:14,876][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:05:16,613][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:05:20,451][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:05:21,563][__main__][INFO] - Number of regex retries in iteration 546: 2 [2026-03-26 08:05:21,563][__main__][INFO] - agents played in iteration 546 are Alice, Bob [2026-03-26 08:05:22,442][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:05:22,486][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:05:22,487][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:05:22,488][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:05:23,095][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:05:23,558][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:05:24,064][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:05:24,566][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:05:25,068][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:05:25,569][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:05:26,069][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:05:26,569][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:05:27,069][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:05:27,569][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:05:28,069][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:05:28,568][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:05:29,068][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:05:29,567][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:05:30,066][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:05:30,564][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:05:31,063][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:05:31,564][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:05:32,062][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:05:32,561][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:05:33,060][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:05:33,559][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:05:34,057][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:05:34,556][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:05:35,055][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:05:35,553][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:05:36,052][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:05:36,551][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:05:37,049][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:05:37,548][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:05:38,047][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:05:38,547][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:05:39,047][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:05:39,547][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:05:40,047][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:05:40,546][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:05:41,046][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:05:41,545][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:05:42,045][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:05:42,545][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:05:43,044][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:05:43,540][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:05:44,034][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:05:44,528][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:05:45,022][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:05:45,520][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:05:46,016][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:05:46,511][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:05:47,006][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:05:47,505][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:05:48,005][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:05:48,507][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:05:49,006][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:05:49,505][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:05:50,004][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:05:50,505][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:05:51,004][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:05:51,506][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:05:52,007][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:05:52,507][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:05:53,007][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:05:53,507][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:05:54,007][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:05:54,507][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:05:55,008][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:05:55,508][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:05:56,007][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:05:56,507][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:05:57,006][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:05:57,506][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:05:58,005][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:05:58,506][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:05:59,006][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:05:59,505][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:06:00,003][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:06:00,501][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:06:01,000][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:06:01,498][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:06:01,996][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:06:02,495][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:06:02,993][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:06:03,493][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:06:03,991][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:06:04,490][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:06:04,989][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:06:05,488][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:06:05,988][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:06:06,487][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:06:06,986][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:06:07,485][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:06:07,984][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:06:08,483][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:06:08,982][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:06:09,482][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:06:09,982][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:06:10,483][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:06:10,983][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:06:11,482][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:06:11,981][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:06:12,480][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:06:12,981][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:06:13,484][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:06:13,985][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:06:14,486][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:06:14,986][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:06:15,482][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:06:15,977][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:06:16,472][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:06:16,968][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:06:17,463][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:06:17,961][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:06:18,461][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:06:18,957][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:06:19,457][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:06:19,956][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:06:20,455][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:06:20,953][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:06:21,452][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:06:21,951][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:06:22,450][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:06:22,950][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:06:23,451][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:06:23,951][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:06:24,450][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:06:24,951][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:06:25,452][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:06:25,952][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:06:26,454][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:06:26,955][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:06:27,587][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 08:06:28,323][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:06:28,326][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:06:28,328][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:06:29,173][__main__][INFO] - Iteration 547 took 1m 14s (9.00% Gen, 89.86% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 23m 4s. Estimated total time: 61h 54m 51s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 49s, 500 more iterations: 10h 19m 8s. [2026-03-26 08:06:29,175][__main__][INFO] - Starting iteration 547. [2026-03-26 08:06:29,574][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:06:29,575][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:06:36,066][__main__][INFO] - Number of regex retries in iteration 547: 0 [2026-03-26 08:06:36,067][__main__][INFO] - agents played in iteration 547 are Alice, Bob [2026-03-26 08:06:36,948][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:06:37,270][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:06:37,271][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:06:37,271][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:06:37,881][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:06:38,337][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:06:38,842][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:06:39,340][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:06:39,840][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:06:40,339][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:06:40,839][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:06:41,338][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:06:41,838][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:06:42,337][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:06:42,835][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:06:43,333][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:06:43,833][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:06:44,331][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:06:44,829][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:06:45,328][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:06:45,828][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:06:46,329][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:06:46,828][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:06:47,327][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:06:47,827][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:06:48,329][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:06:48,829][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:06:49,328][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:06:49,828][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:06:50,326][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:06:50,825][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:06:51,324][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:06:51,823][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:06:52,322][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:06:52,823][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:06:53,322][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:06:53,822][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:06:54,321][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:06:54,820][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:06:55,319][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:06:55,817][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:06:56,316][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:06:56,814][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:06:57,313][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:06:57,812][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:06:58,315][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:06:58,819][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:06:59,321][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:06:59,825][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:07:00,328][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:07:00,830][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:07:01,333][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:07:01,836][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:07:02,334][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:07:02,833][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:07:03,332][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:07:03,831][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:07:04,330][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:07:04,829][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:07:05,328][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:07:05,826][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:07:06,325][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:07:06,824][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:07:07,323][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:07:07,823][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:07:08,322][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:07:08,821][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:07:09,321][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:07:09,820][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:07:10,320][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:07:10,819][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:07:11,318][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:07:11,816][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:07:12,315][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:07:12,814][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:07:13,312][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:07:13,811][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:07:14,311][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:07:14,812][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:07:15,311][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:07:15,812][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:07:16,312][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:07:16,810][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:07:17,309][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:07:17,809][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:07:18,312][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:07:18,813][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:07:19,313][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:07:19,814][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:07:20,314][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:07:20,814][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:07:21,314][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:07:21,815][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:07:22,314][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:07:22,812][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:07:23,312][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:07:23,810][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:07:24,308][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:07:24,807][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:07:25,311][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:07:25,811][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:07:26,310][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:07:26,809][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:07:27,308][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:07:27,807][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:07:28,306][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:07:28,805][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:07:29,304][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:07:29,804][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:07:30,306][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:07:30,809][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:07:31,314][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:07:31,816][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:07:32,318][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:07:32,821][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:07:33,324][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:07:33,827][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:07:34,326][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:07:34,826][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:07:35,325][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:07:35,825][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:07:36,325][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:07:36,825][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:07:37,324][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:07:37,824][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:07:38,323][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:07:38,823][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:07:39,322][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:07:39,822][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:07:40,322][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:07:40,820][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:07:41,318][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:07:41,817][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21751 tokens. [2026-03-26 08:07:42,493][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.57%, ΔTime: 00:01:04 [2026-03-26 08:07:43,226][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:07:43,229][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:07:43,230][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:07:44,076][__main__][INFO] - Iteration 548 took 1m 14s (8.71% Gen, 90.15% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 50h 32m 6s. Estimated total time: 62h 5m 8s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 51s. [2026-03-26 08:07:44,078][__main__][INFO] - Starting iteration 548. [2026-03-26 08:07:44,479][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:07:44,479][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:07:46,178][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:07:51,195][__main__][INFO] - Number of regex retries in iteration 548: 1 [2026-03-26 08:07:51,196][__main__][INFO] - agents played in iteration 548 are Alice, Bob [2026-03-26 08:07:52,073][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:07:52,118][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:07:52,118][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:07:52,119][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:07:52,724][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:07:53,180][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:07:53,684][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:07:54,185][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:07:54,685][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:07:55,187][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:07:55,688][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:07:56,189][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:07:56,690][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:07:57,189][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:07:57,689][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:07:58,188][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:07:58,687][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:07:59,186][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:07:59,685][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:08:00,183][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:08:00,682][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:08:01,180][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:08:01,681][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:08:02,179][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:08:02,678][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:08:03,177][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:08:03,675][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:08:04,174][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:08:04,673][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:08:05,171][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:08:05,670][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:08:06,169][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:08:06,668][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:08:07,166][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:08:07,665][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:08:08,163][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:08:08,663][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:08:09,160][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:08:09,658][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:08:10,155][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:08:10,652][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:08:11,150][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:08:11,649][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:08:12,148][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:08:12,645][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:08:13,144][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:08:13,643][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:08:14,142][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:08:14,642][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:08:15,141][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:08:15,643][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:08:16,143][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:08:16,644][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:08:17,143][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:08:17,641][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:08:18,139][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:08:18,639][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:08:19,138][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:08:19,636][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:08:20,134][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:08:20,632][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:08:21,131][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:08:21,631][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:08:22,129][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:08:22,629][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:08:23,129][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:08:23,629][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:08:24,129][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:08:24,628][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:08:25,130][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:08:25,630][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:08:26,131][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:08:26,632][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:08:27,132][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:08:27,632][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:08:28,134][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:08:28,634][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:08:29,131][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:08:29,629][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:08:30,127][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:08:30,625][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:08:31,124][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:08:31,622][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:08:32,120][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:08:32,618][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:08:33,116][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:08:33,614][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:08:34,112][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:08:34,611][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:08:35,110][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:08:35,608][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:08:36,107][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:08:36,605][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:08:37,103][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:08:37,601][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:08:38,099][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:08:38,599][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:08:39,097][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:08:39,594][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:08:40,092][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:08:40,590][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:08:41,088][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:08:41,586][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:08:42,085][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:08:42,584][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:08:43,082][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:08:43,581][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:08:44,079][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:08:44,577][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:08:45,076][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:08:45,578][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:08:46,080][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:08:46,579][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:08:47,077][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:08:47,577][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:08:48,077][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:08:48,577][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:08:49,076][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:08:49,575][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:08:50,073][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:08:50,572][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:08:51,072][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:08:51,571][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:08:52,070][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:08:52,569][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:08:53,069][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:08:53,570][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:08:54,070][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:08:54,571][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:08:55,072][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:08:55,572][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:08:56,071][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:08:56,571][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 08:08:57,240][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.29%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.49%, ΔTime: 00:01:04 [2026-03-26 08:08:57,978][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:08:57,980][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:08:57,982][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:08:58,631][__main__][INFO] - Iteration 549 took 1m 14s (9.06% Gen, 90.06% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 13m 21s. Estimated total time: 61h 47m 38s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 56s. [2026-03-26 08:08:58,633][__main__][INFO] - Starting iteration 549. [2026-03-26 08:08:59,031][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:08:59,032][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:08:59,624][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:09:05,242][__main__][INFO] - Number of regex retries in iteration 549: 1 [2026-03-26 08:09:05,244][__main__][INFO] - agents played in iteration 549 are Alice, Bob [2026-03-26 08:09:06,126][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:09:06,171][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.52%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:09:06,172][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:09:06,173][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:09:06,780][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:09:07,236][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:09:07,741][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:09:08,240][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:09:08,740][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:09:09,239][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:09:09,737][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:09:10,236][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:09:10,735][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:09:11,236][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:09:11,734][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:09:12,233][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:09:12,732][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:09:13,230][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:09:13,729][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:09:14,229][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:09:14,728][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:09:15,227][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:09:15,726][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:09:16,225][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:09:16,724][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:09:17,224][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:09:17,724][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:09:18,224][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:09:18,727][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:09:19,226][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:09:19,725][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:09:20,223][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:09:20,722][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:09:21,221][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:09:21,720][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:09:22,219][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:09:22,719][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:09:23,218][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:09:23,719][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:09:24,217][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:09:24,717][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:09:25,216][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:09:25,714][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:09:26,213][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:09:26,712][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:09:27,213][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:09:27,712][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:09:28,212][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:09:28,712][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:09:29,212][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:09:29,712][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:09:30,212][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:09:30,711][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:09:31,230][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:09:31,734][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:09:32,234][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:09:32,733][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:09:33,233][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:09:33,732][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:09:34,232][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:09:34,732][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:09:35,232][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:09:35,732][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:09:36,232][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:09:36,731][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:09:37,231][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:09:37,729][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:09:38,229][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:09:38,729][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:09:39,228][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:09:39,726][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:09:40,224][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:09:40,723][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:09:41,221][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:09:41,719][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:09:42,217][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:09:42,715][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:09:43,215][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:09:43,715][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:09:44,214][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:09:44,713][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:09:45,212][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:09:45,711][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:09:46,210][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:09:46,709][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:09:47,209][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:09:47,708][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:09:48,209][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:09:48,709][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:09:49,209][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:09:49,708][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:09:50,207][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:09:50,707][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:09:51,206][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:09:51,704][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:09:52,202][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:09:52,701][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:09:53,199][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:09:53,698][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:09:54,197][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:09:54,695][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:09:55,196][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:09:55,696][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:09:56,194][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:09:56,693][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:09:57,194][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:09:57,692][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:09:58,192][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:09:58,692][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:09:59,194][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:09:59,694][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:10:00,193][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:10:00,694][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:10:01,193][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:10:01,693][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:10:02,198][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:10:02,698][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:10:03,200][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:10:03,701][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:10:04,205][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:10:04,706][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:10:05,208][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:10:05,709][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:10:06,215][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:10:06,717][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:10:07,218][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:10:07,719][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:10:08,220][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:10:08,721][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:10:09,222][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:10:09,728][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:10:10,231][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:10:10,734][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:10:11,403][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.27%, Current % of VRAM taken: 60.63%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 08:10:12,136][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:10:12,139][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:10:12,140][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:10:12,853][__main__][INFO] - Iteration 550 took 1m 13s (8.41% Gen, 90.62% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 49h 55m 37s. Estimated total time: 61h 31m 8s. Time estimates for 10 more iterations: 12m 18s, 100 more iterations: 2h 3m 2s, 500 more iterations: 10h 15m 11s. [2026-03-26 08:10:12,855][__main__][INFO] - Starting iteration 550. [2026-03-26 08:10:13,254][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 10 and human policies 1. [2026-03-26 08:10:13,255][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:10:13,841][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:10:14,821][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:10:19,976][__main__][INFO] - Number of regex retries in iteration 550: 2 [2026-03-26 08:10:19,977][__main__][INFO] - agents played in iteration 550 are Alice, Bob [2026-03-26 08:10:20,865][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:10:20,910][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.50%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:10:20,911][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:10:20,911][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:10:21,521][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:10:21,976][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:10:22,481][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:10:22,981][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:10:23,481][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:10:23,980][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:10:24,479][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:10:24,978][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:10:25,478][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:10:25,982][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:10:26,483][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:10:26,983][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:10:27,483][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:10:27,982][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:10:28,481][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:10:28,982][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:10:29,484][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:10:29,985][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:10:30,487][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:10:30,987][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:10:31,489][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:10:31,990][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:10:32,491][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:10:32,991][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:10:33,493][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:10:33,992][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:10:34,490][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:10:34,990][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:10:35,490][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:10:35,990][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:10:36,489][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:10:36,989][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:10:37,488][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:10:37,988][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:10:38,486][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:10:38,985][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:10:39,485][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:10:39,985][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:10:40,484][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:10:40,983][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:10:41,482][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:10:41,980][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:10:42,479][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:10:42,978][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:10:43,478][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:10:43,978][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:10:44,477][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:10:44,977][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:10:45,476][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:10:45,973][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:10:46,471][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:10:46,970][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:10:47,470][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:10:47,968][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:10:48,468][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:10:48,967][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:10:49,466][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:10:49,965][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:10:50,464][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:10:50,963][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:10:51,465][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:10:51,964][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:10:52,463][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:10:52,960][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:10:53,459][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:10:53,957][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:10:54,455][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:10:54,954][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:10:55,453][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:10:55,952][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:10:56,450][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:10:56,949][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:10:57,449][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:10:57,949][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:10:58,452][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:10:58,954][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:10:59,455][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:10:59,955][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:11:00,455][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:11:00,956][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:11:01,457][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:11:01,959][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:11:02,461][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:11:02,963][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:11:03,465][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:11:03,967][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:11:04,469][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:11:04,970][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:11:05,478][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:11:05,979][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:11:06,479][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:11:06,980][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:11:07,481][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:11:07,981][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:11:08,482][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:11:08,987][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:11:09,489][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:11:09,989][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:11:10,490][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:11:10,991][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:11:11,492][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:11:11,993][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:11:12,493][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:11:12,999][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:11:13,501][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:11:14,003][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:11:14,504][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:11:15,006][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:11:15,507][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:11:16,008][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:11:16,512][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:11:17,015][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:11:17,518][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:11:18,018][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:11:18,522][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:11:19,022][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:11:19,522][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:11:20,029][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:11:20,530][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:11:21,029][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:11:21,530][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:11:22,031][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:11:22,532][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:11:23,032][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:11:23,538][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:11:24,039][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:11:24,540][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:11:25,041][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:11:25,544][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21750 tokens. [2026-03-26 08:11:26,218][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 08:11:26,947][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:11:26,950][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:11:26,951][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:11:28,228][__main__][INFO] - Iteration 551 took 1m 14s (8.96% Gen, 89.33% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 52m 0s. Estimated total time: 62h 28m 46s. Time estimates for 10 more iterations: 12m 29s, 100 more iterations: 2h 4m 57s, 500 more iterations: 10h 24m 47s. [2026-03-26 08:11:28,231][__main__][INFO] - Starting iteration 551. [2026-03-26 08:11:28,631][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:11:28,632][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:11:31,672][mllm.models.large_language_model_local][WARNING] - Response Given the symmetry in per-item values and the previous outcomes where splitting the items equally led to the highest points, I will propose to keep all items again: Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:11:35,512][__main__][INFO] - Number of regex retries in iteration 551: 1 [2026-03-26 08:11:35,513][__main__][INFO] - agents played in iteration 551 are Alice, Bob [2026-03-26 08:11:36,401][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:11:36,446][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:11:36,447][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:11:36,447][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:11:37,059][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:11:37,519][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:11:38,027][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:11:38,531][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:11:39,033][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:11:39,536][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:11:40,038][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:11:40,541][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:11:41,044][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:11:41,542][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:11:42,041][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:11:42,538][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:11:43,038][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:11:43,535][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:11:44,033][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:11:44,532][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:11:45,030][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:11:45,533][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:11:46,035][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:11:46,535][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:11:47,036][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:11:47,535][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:11:48,036][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:11:48,538][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:11:49,038][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:11:49,537][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:11:50,035][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:11:50,535][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:11:51,034][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:11:51,534][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:11:52,033][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:11:52,531][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:11:53,030][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:11:53,529][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:11:54,028][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:11:54,527][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:11:55,027][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:11:55,527][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:11:56,026][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:11:56,526][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:11:57,025][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:11:57,527][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:11:58,028][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:11:58,527][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:11:59,027][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:11:59,527][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:12:00,027][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:12:00,526][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:12:01,026][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:12:01,527][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:12:02,026][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:12:02,526][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:12:03,026][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:12:03,526][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:12:04,027][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:12:04,527][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:12:05,028][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:12:05,527][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:12:06,026][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:12:06,525][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:12:07,024][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:12:07,522][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:12:08,021][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:12:08,521][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:12:09,019][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:12:09,521][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:12:10,023][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:12:10,525][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:12:11,027][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:12:11,530][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:12:12,034][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:12:12,536][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:12:13,039][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:12:13,540][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:12:14,040][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:12:14,538][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:12:15,037][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:12:15,535][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:12:16,035][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:12:16,533][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:12:17,031][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:12:17,531][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:12:18,032][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:12:18,534][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:12:19,034][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:12:19,534][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:12:20,035][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:12:20,535][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:12:21,034][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:12:21,533][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:12:22,033][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:12:22,535][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:12:23,035][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:12:23,535][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:12:24,036][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:12:24,535][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:12:25,035][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:12:25,535][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:12:26,040][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:12:26,541][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:12:27,041][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:12:27,543][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:12:28,044][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:12:28,546][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:12:29,047][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:12:29,549][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:12:30,056][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:12:30,557][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:12:31,058][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:12:31,560][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:12:32,062][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:12:32,564][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:12:33,065][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:12:33,573][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:12:34,076][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:12:34,579][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:12:35,082][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:12:35,584][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:12:36,086][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:12:36,590][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:12:37,094][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:12:37,600][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:12:38,102][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:12:38,604][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:12:39,104][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:12:39,606][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:12:40,108][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:12:40,609][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:12:41,116][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21760 tokens. [2026-03-26 08:12:41,787][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.51%, ΔTime: 00:01:04 [2026-03-26 08:12:42,571][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:12:42,574][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:12:42,575][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:12:43,479][__main__][INFO] - Iteration 552 took 1m 14s (9.19% Gen, 89.60% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 50h 44m 25s. Estimated total time: 62h 22m 26s. Time estimates for 10 more iterations: 12m 28s, 100 more iterations: 2h 4m 44s, 500 more iterations: 10h 23m 44s. [2026-03-26 08:12:43,482][__main__][INFO] - Starting iteration 552. [2026-03-26 08:12:43,880][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:12:43,880][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:12:46,153][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:12:50,454][__main__][INFO] - Number of regex retries in iteration 552: 1 [2026-03-26 08:12:50,455][__main__][INFO] - agents played in iteration 552 are Alice, Bob [2026-03-26 08:12:51,350][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:12:51,395][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:12:51,396][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:12:51,396][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:12:52,008][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:12:52,464][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:12:52,969][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:12:53,470][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:12:53,971][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:12:54,471][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:12:54,971][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:12:55,472][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:12:55,972][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:12:56,471][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:12:56,971][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:12:57,472][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:12:57,972][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:12:58,472][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:12:58,971][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:12:59,471][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:12:59,971][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:13:00,476][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:13:00,976][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:13:01,477][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:13:01,978][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:13:02,477][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:13:02,976][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:13:03,477][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:13:03,977][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:13:04,476][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:13:04,975][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:13:05,475][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:13:05,974][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:13:06,473][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:13:06,972][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:13:07,472][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:13:07,972][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:13:08,471][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:13:08,971][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:13:09,471][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:13:09,971][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:13:10,471][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:13:10,971][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:13:11,472][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:13:11,971][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:13:12,473][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:13:12,973][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:13:13,473][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:13:13,973][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:13:14,474][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:13:14,975][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:13:15,479][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:13:15,981][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:13:16,480][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:13:16,980][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:13:17,479][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:13:17,979][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:13:18,481][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:13:18,980][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:13:19,480][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:13:19,979][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:13:20,476][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:13:20,974][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:13:21,474][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:13:21,971][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:13:22,470][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:13:22,969][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:13:23,468][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:13:23,966][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:13:24,465][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:13:24,967][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:13:25,468][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:13:25,967][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:13:26,469][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:13:26,969][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:13:27,469][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:13:27,969][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:13:28,469][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:13:28,969][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:13:29,469][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:13:29,970][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:13:30,470][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:13:30,971][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:13:31,471][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:13:31,971][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:13:32,472][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:13:32,971][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:13:33,475][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:13:33,976][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:13:34,477][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:13:34,978][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:13:35,478][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:13:35,979][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:13:36,480][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:13:36,981][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:13:37,482][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:13:37,983][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:13:38,485][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:13:38,987][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:13:39,488][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:13:39,988][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:13:40,490][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:13:40,992][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:13:41,494][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:13:41,996][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:13:42,497][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:13:42,999][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:13:43,500][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:13:44,001][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:13:44,508][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:13:45,011][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:13:45,514][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:13:46,017][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:13:46,519][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:13:47,022][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:13:47,524][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:13:48,032][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:13:48,537][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:13:49,039][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:13:49,540][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:13:50,042][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:13:50,544][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:13:51,046][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:13:51,551][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:13:52,055][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:13:52,556][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:13:53,057][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:13:53,558][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:13:54,059][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:13:54,560][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:13:55,065][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:13:55,567][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:13:56,069][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 08:13:56,753][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 08:13:57,545][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:13:57,547][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:13:57,549][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:13:58,245][__main__][INFO] - Iteration 553 took 1m 14s (8.84% Gen, 90.22% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 50h 19m 2s. Estimated total time: 61h 58m 18s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 56s, 500 more iterations: 10h 19m 43s. [2026-03-26 08:13:58,248][__main__][INFO] - Starting iteration 553. [2026-03-26 08:13:58,647][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:13:58,648][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:13:59,836][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:14:05,552][__main__][INFO] - Number of regex retries in iteration 553: 1 [2026-03-26 08:14:05,553][__main__][INFO] - agents played in iteration 553 are Alice, Bob [2026-03-26 08:14:06,459][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:14:06,505][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:14:06,505][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:14:06,506][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:14:07,111][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:14:07,568][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:14:08,074][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:14:08,574][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:14:09,074][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:14:09,575][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:14:10,075][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:14:10,575][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:14:11,075][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:14:11,578][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:14:12,078][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:14:12,580][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:14:13,081][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:14:13,582][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:14:14,083][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:14:14,585][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:14:15,088][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:14:15,587][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:14:16,088][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:14:16,588][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:14:17,088][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:14:17,587][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:14:18,087][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:14:18,589][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:14:19,089][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:14:19,588][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:14:20,088][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:14:20,588][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:14:21,087][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:14:21,586][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:14:22,088][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:14:22,587][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:14:23,089][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:14:23,590][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:14:24,089][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:14:24,588][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:14:25,087][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:14:25,587][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:14:26,086][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:14:26,585][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:14:27,085][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:14:27,585][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:14:28,085][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:14:28,586][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:14:29,086][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:14:29,587][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:14:30,087][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:14:30,588][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:14:31,088][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:14:31,587][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:14:32,086][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:14:32,586][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:14:33,086][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:14:33,585][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:14:34,084][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:14:34,582][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:14:35,081][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:14:35,580][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:14:36,078][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:14:36,578][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:14:37,077][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:14:37,577][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:14:38,076][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:14:38,575][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:14:39,074][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:14:39,573][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:14:40,075][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:14:40,575][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:14:41,076][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:14:41,577][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:14:42,078][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:14:42,578][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:14:43,078][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:14:43,578][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:14:44,078][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:14:44,578][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:14:45,078][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:14:45,581][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:14:46,083][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:14:46,585][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:14:47,087][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:14:47,587][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:14:48,086][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:14:48,587][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:14:49,087][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:14:49,586][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:14:50,087][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:14:50,587][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:14:51,087][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:14:51,586][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:14:52,085][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:14:52,583][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:14:53,082][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:14:53,580][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:14:54,080][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:14:54,579][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:14:55,080][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:14:55,580][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:14:56,079][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:14:56,578][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:14:57,077][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:14:57,577][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:14:58,076][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:14:58,575][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:14:59,075][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:14:59,575][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:15:00,075][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:15:00,574][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:15:01,075][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:15:01,576][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:15:02,076][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:15:02,576][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:15:03,077][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:15:03,575][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:15:04,073][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:15:04,572][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:15:05,072][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:15:05,572][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:15:06,071][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:15:06,572][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:15:07,072][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:15:07,572][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:15:08,072][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:15:08,573][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:15:09,073][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:15:09,572][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:15:10,071][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:15:10,576][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:15:11,077][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 08:15:11,758][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.42%, ΔTime: 00:01:04 [2026-03-26 08:15:12,511][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:15:12,513][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:15:12,515][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:15:13,397][__main__][INFO] - Iteration 554 took 1m 14s (9.24% Gen, 89.58% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 37m 2s. Estimated total time: 62h 17m 33s. Time estimates for 10 more iterations: 12m 27s, 100 more iterations: 2h 4m 35s, 500 more iterations: 10h 22m 55s. [2026-03-26 08:15:13,400][__main__][INFO] - Starting iteration 554. [2026-03-26 08:15:13,798][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:15:13,799][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:15:16,472][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:15:16,526][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:15:20,156][__main__][INFO] - Number of regex retries in iteration 554: 2 [2026-03-26 08:15:20,157][__main__][INFO] - agents played in iteration 554 are Alice, Bob [2026-03-26 08:15:21,036][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:15:21,082][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:15:21,082][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:15:21,083][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:15:21,694][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:15:22,149][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:15:22,655][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:15:23,155][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:15:23,658][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:15:24,158][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:15:24,657][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:15:25,158][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:15:25,657][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:15:26,157][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:15:26,655][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:15:27,154][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:15:27,654][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:15:28,153][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:15:28,652][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:15:29,152][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:15:29,651][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:15:30,151][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:15:30,651][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:15:31,150][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:15:31,650][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:15:32,149][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:15:32,648][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:15:33,147][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:15:33,646][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:15:34,145][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:15:34,644][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:15:35,142][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:15:35,640][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:15:36,139][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:15:36,639][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:15:37,137][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:15:37,637][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:15:38,136][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:15:38,636][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:15:39,135][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:15:39,634][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:15:40,134][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:15:40,635][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:15:41,135][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:15:41,635][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:15:42,134][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:15:42,633][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:15:43,132][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:15:43,632][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:15:44,132][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:15:44,632][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:15:45,131][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:15:45,631][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:15:46,130][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:15:46,629][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:15:47,129][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:15:47,629][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:15:48,129][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:15:48,630][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:15:49,129][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:15:49,628][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:15:50,128][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:15:50,628][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:15:51,126][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:15:51,625][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:15:52,124][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:15:52,623][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:15:53,122][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:15:53,620][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:15:54,120][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:15:54,619][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:15:55,119][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:15:55,618][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:15:56,117][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:15:56,616][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:15:57,115][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:15:57,614][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:15:58,112][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:15:58,611][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:15:59,109][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:15:59,608][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:16:00,106][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:16:00,605][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:16:01,102][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:16:01,600][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:16:02,098][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:16:02,596][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:16:03,095][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:16:03,593][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:16:04,092][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:16:04,591][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:16:05,091][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:16:05,590][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:16:06,089][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:16:06,588][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:16:07,087][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:16:07,587][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:16:08,087][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:16:08,586][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:16:09,084][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:16:09,583][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:16:10,083][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:16:10,584][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:16:11,084][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:16:11,584][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:16:12,084][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:16:12,583][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:16:13,083][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:16:13,582][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:16:14,081][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:16:14,581][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:16:15,081][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:16:15,584][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:16:16,085][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:16:16,584][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:16:17,083][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:16:17,584][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:16:18,085][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:16:18,586][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:16:19,086][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:16:19,585][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:16:20,084][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:16:20,583][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:16:21,083][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:16:21,582][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:16:22,081][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:16:22,581][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:16:23,082][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:16:23,582][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:16:24,082][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:16:24,583][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:16:25,085][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:16:25,585][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:16:26,260][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.21%, Current % of VRAM taken: 60.57%, Block Peak % of device VRAM: 62.32%, ΔTime: 00:01:04 [2026-03-26 08:16:27,001][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:16:27,003][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:16:27,005][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:16:27,841][__main__][INFO] - Iteration 555 took 1m 14s (8.59% Gen, 90.28% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 0m 24s. Estimated total time: 61h 42m 9s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 24s, 500 more iterations: 10h 17m 1s. [2026-03-26 08:16:27,843][__main__][INFO] - Starting iteration 555. [2026-03-26 08:16:28,241][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:16:28,242][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:16:35,196][__main__][INFO] - Number of regex retries in iteration 555: 0 [2026-03-26 08:16:35,197][__main__][INFO] - agents played in iteration 555 are Alice, Bob [2026-03-26 08:16:36,088][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:16:36,132][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.48%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:16:36,133][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:16:36,134][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:16:36,746][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:16:37,203][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:16:37,706][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:16:38,206][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:16:38,704][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:16:39,204][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:16:39,707][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:16:40,207][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:16:40,706][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:16:41,207][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:16:41,707][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:16:42,206][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:16:42,705][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:16:43,207][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:16:43,707][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:16:44,206][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:16:44,705][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:16:45,205][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:16:45,703][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:16:46,201][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:16:46,700][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:16:47,199][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:16:47,698][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:16:48,200][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:16:48,699][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:16:49,198][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:16:49,698][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:16:50,198][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:16:50,697][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:16:51,196][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:16:51,696][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:16:52,195][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:16:52,695][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:16:53,194][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:16:53,693][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:16:54,193][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:16:54,693][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:16:55,193][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:16:55,693][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:16:56,192][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:16:56,692][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:16:57,191][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:16:57,690][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:16:58,189][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:16:58,689][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:16:59,188][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:16:59,687][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:17:00,185][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:17:00,684][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:17:01,183][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:17:01,683][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:17:02,183][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:17:02,682][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:17:03,182][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:17:03,681][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:17:04,180][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:17:04,678][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:17:05,177][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:17:05,676][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:17:06,175][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:17:06,675][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:17:07,173][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:17:07,672][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:17:08,170][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:17:08,669][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:17:09,168][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:17:09,668][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:17:10,167][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:17:10,667][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:17:11,166][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:17:11,664][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:17:12,163][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:17:12,661][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:17:13,161][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:17:13,660][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:17:14,159][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:17:14,658][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:17:15,157][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:17:15,656][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:17:16,155][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:17:16,654][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:17:17,153][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:17:17,652][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:17:18,152][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:17:18,654][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:17:19,153][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:17:19,652][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:17:20,151][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:17:20,651][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:17:21,150][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:17:21,650][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:17:22,149][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:17:22,649][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:17:23,149][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:17:23,649][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:17:24,148][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:17:24,649][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:17:25,149][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:17:25,650][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:17:26,150][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:17:26,650][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:17:27,149][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:17:27,649][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:17:28,149][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:17:28,648][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:17:29,147][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:17:29,647][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:17:30,146][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:17:30,645][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:17:31,144][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:17:31,643][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:17:32,142][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:17:32,640][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:17:33,139][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:17:33,638][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:17:34,138][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:17:34,636][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:17:35,134][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:17:35,634][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:17:36,134][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:17:36,634][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:17:37,132][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:17:37,631][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:17:38,129][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:17:38,628][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:17:39,125][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:17:39,624][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:17:40,121][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:17:40,619][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21755 tokens. [2026-03-26 08:17:41,290][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.51%, Block Peak % of device VRAM: 62.10%, ΔTime: 00:01:04 [2026-03-26 08:17:42,027][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:17:42,030][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:17:42,032][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:17:42,683][__main__][INFO] - Iteration 556 took 1m 14s (9.34% Gen, 89.78% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 19m 5s. Estimated total time: 62h 2m 6s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 4s, 500 more iterations: 10h 20m 21s. [2026-03-26 08:17:42,685][__main__][INFO] - Starting iteration 556. [2026-03-26 08:17:43,083][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:17:43,084][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:17:49,838][__main__][INFO] - Number of regex retries in iteration 556: 0 [2026-03-26 08:17:49,839][__main__][INFO] - agents played in iteration 556 are Alice, Bob [2026-03-26 08:17:50,717][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:17:50,763][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:17:50,763][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:17:50,764][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:17:51,373][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:17:51,829][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:17:52,333][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:17:52,833][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:17:53,332][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:17:53,832][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:17:54,332][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:17:54,833][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:17:55,333][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:17:55,833][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:17:56,332][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:17:56,832][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:17:57,332][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:17:57,832][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:17:58,332][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:17:58,831][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:17:59,330][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:17:59,829][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:18:00,328][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:18:00,827][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:18:01,328][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:18:01,827][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:18:02,327][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:18:02,826][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:18:03,325][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:18:03,825][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:18:04,324][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:18:04,824][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:18:05,324][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:18:05,824][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:18:06,323][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:18:06,823][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:18:07,324][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:18:07,825][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:18:08,325][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:18:08,825][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:18:09,324][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:18:09,823][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:18:10,320][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:18:10,820][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:18:11,318][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:18:11,816][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:18:12,315][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:18:12,813][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:18:13,311][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:18:13,810][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:18:14,308][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:18:14,805][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:18:15,303][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:18:15,800][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:18:16,298][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:18:16,795][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:18:17,293][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:18:17,792][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:18:18,292][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:18:18,791][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:18:19,288][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:18:19,787][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:18:20,286][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:18:20,787][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:18:21,288][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:18:21,788][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:18:22,287][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:18:22,787][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:18:23,286][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:18:23,786][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:18:24,286][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:18:24,787][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:18:25,289][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:18:25,789][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:18:26,288][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:18:26,787][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:18:27,287][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:18:27,786][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:18:28,285][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:18:28,784][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:18:29,283][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:18:29,781][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:18:30,279][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:18:30,777][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:18:31,275][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:18:31,774][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:18:32,273][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:18:32,771][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:18:33,270][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:18:33,768][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:18:34,266][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:18:34,765][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:18:35,263][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:18:35,763][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:18:36,263][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:18:36,762][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:18:37,261][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:18:37,761][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:18:38,260][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:18:38,758][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:18:39,257][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:18:39,756][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:18:40,256][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:18:40,755][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:18:41,254][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:18:41,753][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:18:42,253][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:18:42,752][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:18:43,252][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:18:43,752][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:18:44,251][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:18:44,750][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:18:45,248][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:18:45,745][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:18:46,243][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:18:46,740][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:18:47,238][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:18:47,736][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:18:48,236][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:18:48,735][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:18:49,233][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:18:49,731][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:18:50,230][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:18:50,728][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:18:51,226][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:18:51,729][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:18:52,229][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:18:52,728][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:18:53,227][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:18:53,727][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:18:54,225][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:18:54,725][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:18:55,227][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 08:18:55,861][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.25%, Current % of VRAM taken: 60.61%, Block Peak % of device VRAM: 62.28%, ΔTime: 00:01:04 [2026-03-26 08:18:56,644][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:18:56,646][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:18:56,648][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:18:57,330][__main__][INFO] - Iteration 557 took 1m 14s (9.10% Gen, 89.98% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 8m 6s. Estimated total time: 61h 52m 22s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 44s, 500 more iterations: 10h 18m 43s. [2026-03-26 08:18:57,332][__main__][INFO] - Starting iteration 557. [2026-03-26 08:18:57,733][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:18:57,734][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:18:58,770][mllm.models.large_language_model_local][WARNING] - Response Proposal: 0 hats, 0 books, 30 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:19:04,150][__main__][INFO] - Number of regex retries in iteration 557: 1 [2026-03-26 08:19:04,151][__main__][INFO] - agents played in iteration 557 are Alice, Bob [2026-03-26 08:19:05,031][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:19:05,076][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:19:05,076][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:19:05,077][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:19:05,685][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:19:06,142][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:19:06,647][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:19:07,148][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:19:07,650][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:19:08,151][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:19:08,653][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:19:09,154][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:19:09,655][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:19:10,155][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:19:10,654][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:19:11,154][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:19:11,653][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:19:12,153][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:19:12,652][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:19:13,152][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:19:13,652][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:19:14,152][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:19:14,653][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:19:15,153][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:19:15,656][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:19:16,158][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:19:16,658][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:19:17,158][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:19:17,658][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:19:18,158][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:19:18,656][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:19:19,155][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:19:19,653][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:19:20,153][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:19:20,652][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:19:21,150][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:19:21,648][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:19:22,149][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:19:22,651][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:19:23,151][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:19:23,653][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:19:24,154][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:19:24,655][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:19:25,157][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:19:25,658][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:19:26,157][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:19:26,658][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:19:27,158][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:19:27,659][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:19:28,160][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:19:28,660][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:19:29,160][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:19:29,660][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:19:30,159][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:19:30,659][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:19:31,160][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:19:31,660][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:19:32,160][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:19:32,660][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:19:33,159][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:19:33,659][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:19:34,159][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:19:34,659][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:19:35,157][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:19:35,656][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:19:36,154][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:19:36,654][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:19:37,153][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:19:37,653][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:19:38,154][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:19:38,656][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:19:39,156][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:19:39,656][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:19:40,156][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:19:40,657][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:19:41,158][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:19:41,659][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:19:42,156][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:19:42,654][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:19:43,153][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:19:43,651][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:19:44,149][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:19:44,649][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:19:45,148][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:19:45,646][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:19:46,146][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:19:46,645][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:19:47,145][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:19:47,644][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:19:48,146][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:19:48,650][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:19:49,151][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:19:49,651][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:19:50,150][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:19:50,649][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:19:51,147][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:19:51,647][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:19:52,146][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:19:52,645][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:19:53,144][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:19:53,642][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:19:54,142][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:19:54,643][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:19:55,143][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:19:55,644][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:19:56,144][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:19:56,645][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:19:57,144][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:19:57,645][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:19:58,145][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:19:58,646][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:19:59,146][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:19:59,647][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:20:00,147][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:20:00,647][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:20:01,148][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:20:01,648][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:20:02,150][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:20:02,650][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:20:03,151][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:20:03,651][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:20:04,152][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:20:04,653][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:20:05,154][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:20:05,654][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:20:06,155][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:20:06,654][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:20:07,152][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:20:07,651][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:20:08,151][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:20:08,650][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:20:09,149][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:20:09,649][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:20:10,279][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.60%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 08:20:11,061][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:20:11,063][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:20:11,065][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:20:11,759][__main__][INFO] - Iteration 558 took 1m 14s (8.67% Gen, 90.39% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 49h 55m 48s. Estimated total time: 61h 41m 18s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 22s, 500 more iterations: 10h 16m 53s. [2026-03-26 08:20:11,761][__main__][INFO] - Starting iteration 558. [2026-03-26 08:20:12,159][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:20:12,160][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:20:19,217][__main__][INFO] - Number of regex retries in iteration 558: 0 [2026-03-26 08:20:19,218][__main__][INFO] - agents played in iteration 558 are Alice, Bob [2026-03-26 08:20:20,092][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:20:20,137][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:20:20,137][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:20:20,138][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:20:20,746][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:20:21,200][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:20:21,703][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:20:22,202][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:20:22,701][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:20:23,199][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:20:23,699][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:20:24,198][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:20:24,696][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:20:25,196][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:20:25,696][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:20:26,195][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:20:26,694][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:20:27,193][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:20:27,692][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:20:28,191][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:20:28,689][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:20:29,190][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:20:29,691][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:20:30,190][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:20:30,690][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:20:31,189][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:20:31,689][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:20:32,188][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:20:32,688][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:20:33,186][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:20:33,684][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:20:34,182][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:20:34,679][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:20:35,176][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:20:35,675][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:20:36,173][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:20:36,672][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:20:37,171][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:20:37,669][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:20:38,168][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:20:38,666][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:20:39,164][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:20:39,662][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:20:40,160][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:20:40,658][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:20:41,159][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:20:41,659][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:20:42,159][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:20:42,657][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:20:43,157][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:20:43,656][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:20:44,157][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:20:44,658][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:20:45,156][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:20:45,659][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:20:46,158][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:20:46,656][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:20:47,156][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:20:47,655][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:20:48,154][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:20:48,653][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:20:49,153][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:20:49,653][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:20:50,152][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:20:50,652][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:20:51,150][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:20:51,649][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:20:52,149][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:20:52,647][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:20:53,144][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:20:53,643][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:20:54,140][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:20:54,638][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:20:55,137][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:20:55,635][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:20:56,133][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:20:56,631][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:20:57,128][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:20:57,626][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:20:58,124][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:20:58,621][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:20:59,118][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:20:59,616][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:21:00,113][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:21:00,612][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:21:01,111][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:21:01,609][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:21:02,108][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:21:02,607][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:21:03,106][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:21:03,606][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:21:04,107][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:21:04,606][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:21:05,104][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:21:05,602][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:21:06,100][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:21:06,597][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:21:07,095][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:21:07,593][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:21:08,091][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:21:08,590][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:21:09,088][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:21:09,586][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:21:10,084][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:21:10,582][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:21:11,080][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:21:11,578][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:21:12,078][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:21:12,577][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:21:13,078][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:21:13,579][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:21:14,081][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:21:14,581][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:21:15,081][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:21:15,581][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:21:16,081][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:21:16,581][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:21:17,079][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:21:17,577][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:21:18,076][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:21:18,577][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:21:19,075][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:21:19,574][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:21:20,074][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:21:20,573][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:21:21,071][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:21:21,570][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:21:22,069][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:21:22,569][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:21:23,068][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:21:23,569][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:21:24,068][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:21:24,567][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21754 tokens. [2026-03-26 08:21:25,212][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.18%, Current % of VRAM taken: 60.54%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 08:21:25,948][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:21:25,950][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:21:25,952][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:21:26,659][__main__][INFO] - Iteration 559 took 1m 14s (9.47% Gen, 89.58% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 18m 15s. Estimated total time: 62h 5m 0s. Time estimates for 10 more iterations: 12m 25s, 100 more iterations: 2h 4m 10s, 500 more iterations: 10h 20m 50s. [2026-03-26 08:21:26,661][__main__][INFO] - Starting iteration 559. [2026-03-26 08:21:27,062][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:21:27,062][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:21:27,632][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 0 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:21:28,770][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:21:34,349][__main__][INFO] - Number of regex retries in iteration 559: 2 [2026-03-26 08:21:34,350][__main__][INFO] - agents played in iteration 559 are Alice, Bob [2026-03-26 08:21:35,230][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:21:35,551][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:21:35,552][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:21:35,553][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:21:36,157][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:21:36,612][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:21:37,115][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:21:37,614][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:21:38,114][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:21:38,613][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:21:39,112][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:21:39,611][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:21:40,110][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:21:40,617][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:21:41,117][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:21:41,617][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:21:42,117][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:21:42,618][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:21:43,118][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:21:43,618][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:21:44,118][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:21:44,617][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:21:45,115][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:21:45,614][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:21:46,113][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:21:46,612][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:21:47,111][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:21:47,609][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:21:48,108][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:21:48,610][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:21:49,110][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:21:49,609][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:21:50,108][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:21:50,606][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:21:51,105][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:21:51,604][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:21:52,102][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:21:52,602][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:21:53,102][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:21:53,600][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:21:54,100][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:21:54,600][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:21:55,100][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:21:55,599][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:21:56,099][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:21:56,599][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:21:57,098][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:21:57,598][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:21:58,098][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:21:58,597][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:21:59,096][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:21:59,595][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:22:00,096][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:22:00,597][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:22:01,097][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:22:01,597][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:22:02,097][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:22:02,596][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:22:03,096][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:22:03,596][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:22:04,098][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:22:04,597][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:22:05,095][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:22:05,593][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:22:06,091][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:22:06,590][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:22:07,088][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:22:07,587][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:22:08,087][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:22:08,587][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:22:09,086][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:22:09,584][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:22:10,083][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:22:10,582][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:22:11,080][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:22:11,579][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:22:12,078][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:22:12,578][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:22:13,078][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:22:13,577][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:22:14,077][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:22:14,577][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:22:15,078][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:22:15,581][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:22:16,085][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:22:16,584][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:22:17,082][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:22:17,581][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:22:18,081][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:22:18,582][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:22:19,080][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:22:19,580][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:22:20,079][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:22:20,578][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:22:21,077][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:22:21,575][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:22:22,076][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:22:22,575][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:22:23,075][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:22:23,574][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:22:24,074][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:22:24,574][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:22:25,075][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:22:25,575][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:22:26,076][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:22:26,577][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:22:27,077][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:22:27,577][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:22:28,077][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:22:28,577][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:22:29,077][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:22:29,577][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:22:30,078][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:22:30,579][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:22:31,079][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:22:31,578][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:22:32,077][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:22:32,578][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:22:33,079][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:22:33,579][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:22:34,079][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:22:34,579][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:22:35,078][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:22:35,577][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:22:36,079][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:22:36,578][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:22:37,078][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:22:37,576][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:22:38,075][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:22:38,574][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:22:39,073][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:22:39,571][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:22:40,071][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21752 tokens. [2026-03-26 08:22:40,704][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.17%, Current % of VRAM taken: 60.53%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 08:22:41,432][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:22:41,434][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:22:41,438][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:22:42,077][__main__][INFO] - Iteration 560 took 1m 15s (9.71% Gen, 89.43% Train). Generation: 7s, Training: 1m 7s. Estimated remaining time: 50h 42m 49s. Estimated total time: 62h 30m 49s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 1s, 500 more iterations: 10h 25m 8s. [2026-03-26 08:22:42,080][__main__][INFO] - Starting iteration 560. [2026-03-26 08:22:42,483][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:22:42,483][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:22:45,055][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:22:48,976][__main__][INFO] - Number of regex retries in iteration 560: 1 [2026-03-26 08:22:48,977][__main__][INFO] - agents played in iteration 560 are Alice, Bob [2026-03-26 08:22:49,855][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:22:49,900][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:22:49,900][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:22:49,901][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:22:50,507][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:22:50,964][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:22:51,469][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:22:51,971][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:22:52,472][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:22:52,973][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:22:53,474][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:22:53,975][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:22:54,476][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:22:54,976][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:22:55,475][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:22:55,974][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:22:56,474][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:22:56,973][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:22:57,472][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:22:57,971][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:22:58,470][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:22:58,971][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:22:59,472][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:22:59,972][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:23:00,473][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:23:00,973][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:23:01,473][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:23:01,973][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:23:02,473][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:23:02,973][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:23:03,472][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:23:03,972][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:23:04,472][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:23:04,973][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:23:05,471][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:23:05,972][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:23:06,472][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:23:06,973][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:23:07,474][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:23:07,974][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:23:08,474][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:23:08,975][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:23:09,476][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:23:09,977][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:23:10,478][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:23:10,978][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:23:11,477][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:23:11,976][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:23:12,476][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:23:12,976][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:23:13,475][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:23:13,974][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:23:14,474][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:23:14,975][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:23:15,475][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:23:15,975][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:23:16,477][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:23:16,978][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:23:17,478][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:23:17,978][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:23:18,481][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:23:18,980][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:23:19,480][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:23:19,979][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:23:20,479][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:23:20,978][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:23:21,477][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:23:21,976][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:23:22,476][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:23:22,976][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:23:23,477][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:23:23,978][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:23:24,478][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:23:24,979][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:23:25,479][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:23:25,979][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:23:26,479][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:23:26,978][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:23:27,477][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:23:27,975][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:23:28,475][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:23:28,973][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:23:29,473][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:23:29,971][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:23:30,471][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:23:30,971][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:23:31,471][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:23:31,973][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:23:32,474][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:23:32,974][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:23:33,475][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:23:33,975][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:23:34,476][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:23:34,975][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:23:35,475][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:23:35,975][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:23:36,475][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:23:36,975][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:23:37,475][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:23:37,975][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:23:38,475][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:23:38,977][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:23:39,477][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:23:39,977][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:23:40,477][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:23:40,977][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:23:41,478][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:23:41,978][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:23:42,477][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:23:42,976][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:23:43,476][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:23:43,977][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:23:44,477][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:23:44,977][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:23:45,480][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:23:45,981][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:23:46,481][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:23:46,982][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:23:47,482][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:23:47,982][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:23:48,485][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:23:48,985][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:23:49,484][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:23:49,984][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:23:50,485][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:23:50,985][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:23:51,484][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:23:51,984][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:23:52,483][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:23:52,982][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:23:53,480][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:23:53,979][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:23:54,479][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:23:55,113][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.26%, Current % of VRAM taken: 60.62%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 08:23:55,851][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:23:55,853][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:23:55,855][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:23:56,590][__main__][INFO] - Iteration 561 took 1m 14s (8.76% Gen, 90.24% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 49h 56m 12s. Estimated total time: 61h 45m 26s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 30s, 500 more iterations: 10h 17m 34s. [2026-03-26 08:23:56,592][__main__][INFO] - Starting iteration 561. [2026-03-26 08:23:56,995][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:23:56,996][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:23:59,211][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:24:03,739][__main__][INFO] - Number of regex retries in iteration 561: 1 [2026-03-26 08:24:03,740][__main__][INFO] - agents played in iteration 561 are Alice, Bob [2026-03-26 08:24:04,614][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:24:04,659][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:24:04,660][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:24:04,661][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:24:05,268][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:24:05,722][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:24:06,226][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:24:06,725][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:24:07,224][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:24:07,723][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:24:08,221][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:24:08,720][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:24:09,218][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:24:09,719][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:24:10,219][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:24:10,721][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:24:11,223][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:24:11,724][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:24:12,223][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:24:12,724][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:24:13,223][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:24:13,722][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:24:14,220][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:24:14,719][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:24:15,218][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:24:15,717][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:24:16,215][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:24:16,714][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:24:17,213][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:24:17,713][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:24:18,215][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:24:18,715][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:24:19,215][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:24:19,715][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:24:20,214][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:24:20,714][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:24:21,213][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:24:21,716][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:24:22,216][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:24:22,716][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:24:23,216][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:24:23,717][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:24:24,217][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:24:24,718][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:24:25,219][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:24:25,719][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:24:26,218][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:24:26,717][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:24:27,216][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:24:27,716][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:24:28,216][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:24:28,715][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:24:29,215][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:24:29,715][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:24:30,216][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:24:30,715][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:24:31,216][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:24:31,716][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:24:32,216][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:24:32,716][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:24:33,218][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:24:33,719][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:24:34,218][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:24:34,717][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:24:35,217][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:24:35,717][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:24:36,217][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:24:36,717][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:24:37,216][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:24:37,715][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:24:38,214][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:24:38,713][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:24:39,212][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:24:39,712][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:24:40,211][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:24:40,709][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:24:41,208][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:24:41,707][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:24:42,205][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:24:42,704][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:24:43,204][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:24:43,704][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:24:44,203][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:24:44,703][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:24:45,202][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:24:45,701][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:24:46,200][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:24:46,698][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:24:47,197][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:24:47,696][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:24:48,197][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:24:48,697][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:24:49,196][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:24:49,696][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:24:50,196][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:24:50,695][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:24:51,194][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:24:51,695][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:24:52,195][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:24:52,694][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:24:53,194][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:24:53,695][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:24:54,195][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:24:54,695][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:24:55,196][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:24:55,697][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:24:56,197][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:24:56,697][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:24:57,197][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:24:57,695][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:24:58,195][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:24:58,694][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:24:59,192][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:24:59,690][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:25:00,188][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:25:00,685][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:25:01,184][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:25:01,684][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:25:02,184][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:25:02,684][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:25:03,184][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:25:03,684][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:25:04,184][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:25:04,684][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:25:05,184][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:25:05,684][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:25:06,183][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:25:06,681][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:25:07,180][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:25:07,677][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:25:08,176][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:25:08,676][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:25:09,175][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21756 tokens. [2026-03-26 08:25:09,819][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.19%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 08:25:10,554][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:25:10,556][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:25:10,558][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:25:11,213][__main__][INFO] - Iteration 562 took 1m 14s (9.09% Gen, 90.03% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 0m 26s. Estimated total time: 61h 50m 55s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 41s, 500 more iterations: 10h 18m 29s. [2026-03-26 08:25:11,215][__main__][INFO] - Starting iteration 562. [2026-03-26 08:25:11,615][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:25:11,616][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:25:18,372][__main__][INFO] - Number of regex retries in iteration 562: 0 [2026-03-26 08:25:18,373][__main__][INFO] - agents played in iteration 562 are Alice, Bob [2026-03-26 08:25:19,259][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:25:19,304][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:25:19,305][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:25:19,305][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:25:19,917][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:25:20,372][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:25:20,877][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:25:21,376][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:25:21,876][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:25:22,376][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:25:22,875][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:25:23,375][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:25:23,875][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:25:24,372][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:25:24,870][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:25:25,368][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:25:25,866][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:25:26,363][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:25:26,860][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:25:27,357][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:25:27,855][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:25:28,356][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:25:28,855][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:25:29,353][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:25:29,852][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:25:30,352][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:25:30,852][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:25:31,351][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:25:31,850][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:25:32,352][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:25:32,852][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:25:33,352][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:25:33,853][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:25:34,353][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:25:34,853][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:25:35,353][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:25:35,854][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:25:36,354][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:25:36,853][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:25:37,353][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:25:37,853][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:25:38,352][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:25:38,853][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:25:39,352][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:25:39,852][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:25:40,351][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:25:40,852][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:25:41,352][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:25:41,853][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:25:42,353][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:25:42,854][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:25:43,354][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:25:43,853][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:25:44,353][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:25:44,852][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:25:45,351][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:25:45,851][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:25:46,350][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:25:46,849][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:25:47,348][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:25:47,847][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:25:48,349][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:25:48,849][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:25:49,349][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:25:49,851][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:25:50,351][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:25:50,851][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:25:51,350][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:25:51,851][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:25:52,350][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:25:52,851][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:25:53,351][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:25:53,850][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:25:54,350][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:25:54,848][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:25:55,347][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:25:55,846][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:25:56,342][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:25:56,840][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:25:57,338][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:25:57,837][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:25:58,335][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:25:58,832][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:25:59,328][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:25:59,826][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:26:00,324][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:26:00,823][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:26:01,322][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:26:01,821][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:26:02,321][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:26:02,819][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:26:03,318][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:26:03,819][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:26:04,319][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:26:04,819][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:26:05,320][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:26:05,820][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:26:06,321][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:26:06,821][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:26:07,321][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:26:07,821][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:26:08,320][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:26:08,820][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:26:09,319][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:26:09,819][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:26:10,318][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:26:10,817][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:26:11,316][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:26:11,816][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:26:12,316][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:26:12,817][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:26:13,317][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:26:13,818][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:26:14,317][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:26:14,817][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:26:15,317][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:26:15,817][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:26:16,316][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:26:16,815][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:26:17,314][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:26:17,813][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:26:18,314][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:26:18,814][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:26:19,313][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:26:19,813][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:26:20,313][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:26:20,812][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:26:21,311][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:26:21,812][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:26:22,311][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:26:22,812][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:26:23,311][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:26:23,812][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21745 tokens. [2026-03-26 08:26:24,463][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.24%, Current % of VRAM taken: 60.59%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 08:26:25,196][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:26:25,199][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:26:25,200][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:26:25,849][__main__][INFO] - Iteration 563 took 1m 14s (9.10% Gen, 90.02% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 50h 0m 1s. Estimated total time: 61h 51m 45s. Time estimates for 10 more iterations: 12m 22s, 100 more iterations: 2h 3m 43s, 500 more iterations: 10h 18m 37s. [2026-03-26 08:26:25,852][__main__][INFO] - Starting iteration 563. [2026-03-26 08:26:26,256][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:26:26,257][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:26:26,850][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:26:26,855][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:26:33,306][__main__][INFO] - Number of regex retries in iteration 563: 2 [2026-03-26 08:26:33,306][__main__][INFO] - agents played in iteration 563 are Alice, Bob [2026-03-26 08:26:34,185][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:26:34,230][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.49%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:26:34,231][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:26:34,231][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:26:34,842][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:26:35,296][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:26:35,799][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:26:36,299][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:26:36,797][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:26:37,296][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:26:37,796][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:26:38,294][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:26:38,793][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:26:39,294][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:26:39,792][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:26:40,290][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:26:40,788][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:26:41,288][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:26:41,787][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:26:42,286][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:26:42,784][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:26:43,287][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:26:43,790][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:26:44,292][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:26:44,795][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:26:45,299][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:26:45,805][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:26:46,308][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:26:46,810][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:26:47,311][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:26:47,811][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:26:48,312][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:26:48,811][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:26:49,310][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:26:49,808][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:26:50,306][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:26:50,804][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:26:51,303][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:26:51,802][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:26:52,302][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:26:52,801][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:26:53,300][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:26:53,799][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:26:54,298][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:26:54,798][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:26:55,301][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:26:55,801][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:26:56,301][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:26:56,802][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:26:57,303][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:26:57,803][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:26:58,303][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:26:58,802][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:26:59,298][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:26:59,794][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:27:00,289][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:27:00,784][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:27:01,280][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:27:01,776][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:27:02,273][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:27:02,771][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:27:03,271][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:27:03,770][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:27:04,268][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:27:04,767][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:27:05,264][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:27:05,762][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:27:06,259][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:27:06,757][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:27:07,256][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:27:07,755][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:27:08,253][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:27:08,752][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:27:09,249][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:27:09,747][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:27:10,245][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:27:10,742][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:27:11,240][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:27:11,739][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:27:12,239][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:27:12,739][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:27:13,238][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:27:13,737][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:27:14,239][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:27:14,739][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:27:15,241][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:27:15,743][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:27:16,246][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:27:16,748][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:27:17,250][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:27:17,752][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:27:18,258][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:27:18,761][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:27:19,260][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:27:19,758][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:27:20,258][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:27:20,756][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:27:21,255][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:27:21,753][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:27:22,252][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:27:22,751][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:27:23,250][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:27:23,749][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:27:24,248][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:27:24,747][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:27:25,246][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:27:25,746][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:27:26,245][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:27:26,745][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:27:27,244][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:27:27,744][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:27:28,244][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:27:28,743][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:27:29,242][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:27:29,742][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:27:30,243][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:27:30,742][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:27:31,237][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:27:31,732][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:27:32,227][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:27:32,723][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:27:33,217][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:27:33,711][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:27:34,208][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:27:34,703][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:27:35,202][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:27:35,700][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:27:36,198][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:27:36,697][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:27:37,195][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:27:37,694][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:27:38,193][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:27:38,691][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 08:27:39,332][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.15%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.57%, ΔTime: 00:01:04 [2026-03-26 08:27:40,069][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:27:40,071][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:27:40,073][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:27:40,721][__main__][INFO] - Iteration 564 took 1m 14s (9.47% Gen, 89.66% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 10m 16s. Estimated total time: 62h 3m 14s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 6s, 500 more iterations: 10h 20m 32s. [2026-03-26 08:27:40,723][__main__][INFO] - Starting iteration 564. [2026-03-26 08:27:41,128][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:27:41,128][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:27:47,710][__main__][INFO] - Number of regex retries in iteration 564: 0 [2026-03-26 08:27:47,711][__main__][INFO] - agents played in iteration 564 are Alice, Bob [2026-03-26 08:27:48,595][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:27:48,640][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.44%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:27:48,641][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:27:48,641][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:27:49,245][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:27:49,702][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:27:50,207][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:27:50,708][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:27:51,208][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:27:51,710][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:27:52,211][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:27:52,712][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:27:53,212][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:27:53,711][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:27:54,210][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:27:54,708][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:27:55,207][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:27:55,706][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:27:56,204][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:27:56,703][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:27:57,202][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:27:57,701][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:27:58,200][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:27:58,700][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:27:59,199][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:27:59,698][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:28:00,198][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:28:00,697][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:28:01,196][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:28:01,696][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:28:02,196][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:28:02,696][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:28:03,197][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:28:03,696][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:28:04,196][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:28:04,695][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:28:05,194][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:28:05,691][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:28:06,190][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:28:06,690][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:28:07,189][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:28:07,687][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:28:08,186][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:28:08,683][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:28:09,181][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:28:09,680][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:28:10,180][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:28:10,680][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:28:11,180][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:28:11,681][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:28:12,180][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:28:12,680][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:28:13,179][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:28:13,678][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:28:14,176][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:28:14,675][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:28:15,174][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:28:15,676][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:28:16,176][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:28:16,676][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:28:17,175][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:28:17,675][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:28:18,178][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:28:18,679][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:28:19,179][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:28:19,679][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:28:20,179][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:28:20,679][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:28:21,179][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:28:21,681][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:28:22,182][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:28:22,683][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:28:23,183][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:28:23,683][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:28:24,183][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:28:24,684][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:28:25,185][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:28:25,684][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:28:26,182][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:28:26,681][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:28:27,180][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:28:27,678][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:28:28,178][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:28:28,677][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:28:29,176][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:28:29,676][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:28:30,176][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:28:30,676][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:28:31,176][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:28:31,675][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:28:32,175][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:28:32,675][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:28:33,174][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:28:33,673][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:28:34,173][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:28:34,673][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:28:35,173][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:28:35,672][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:28:36,171][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:28:36,671][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:28:37,170][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:28:37,668][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:28:38,167][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:28:38,665][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:28:39,163][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:28:39,660][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:28:40,159][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:28:40,658][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:28:41,156][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:28:41,656][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:28:42,156][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:28:42,656][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:28:43,155][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:28:43,655][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:28:44,155][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:28:44,655][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:28:45,155][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:28:45,654][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:28:46,154][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:28:46,653][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:28:47,152][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:28:47,651][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:28:48,151][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:28:48,651][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:28:49,150][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:28:49,649][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:28:50,149][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:28:50,648][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:28:51,146][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:28:51,646][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:28:52,145][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:28:52,644][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:28:53,143][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 08:28:53,770][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.28%, Current % of VRAM taken: 60.64%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 08:28:54,512][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:28:54,514][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:28:54,516][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:28:55,172][__main__][INFO] - Iteration 565 took 1m 14s (8.89% Gen, 90.22% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 49h 48m 0s. Estimated total time: 61h 42m 13s. Time estimates for 10 more iterations: 12m 20s, 100 more iterations: 2h 3m 24s, 500 more iterations: 10h 17m 2s. [2026-03-26 08:28:55,174][__main__][INFO] - Starting iteration 565. [2026-03-26 08:28:55,575][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:28:55,575][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:28:57,302][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:28:58,566][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:29:02,605][__main__][INFO] - Number of regex retries in iteration 565: 2 [2026-03-26 08:29:02,606][__main__][INFO] - agents played in iteration 565 are Alice, Bob [2026-03-26 08:29:03,482][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:29:03,526][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:29:03,527][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:29:03,528][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:29:04,130][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:29:04,585][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:29:05,090][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:29:05,590][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:29:06,090][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:29:06,589][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:29:07,087][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:29:07,586][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:29:08,086][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:29:08,587][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:29:09,088][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:29:09,589][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:29:10,089][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:29:10,588][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:29:11,089][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:29:11,589][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:29:12,090][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:29:12,590][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:29:13,115][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:29:13,614][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:29:14,115][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:29:14,614][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:29:15,113][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:29:15,612][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:29:16,113][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:29:16,612][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:29:17,111][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:29:17,611][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:29:18,110][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:29:18,612][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:29:19,110][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:29:19,609][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:29:20,107][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:29:20,605][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:29:21,104][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:29:21,603][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:29:22,102][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:29:22,602][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:29:23,101][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:29:23,600][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:29:24,099][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:29:24,599][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:29:25,098][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:29:25,597][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:29:26,096][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:29:26,596][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:29:27,095][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:29:27,593][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:29:28,092][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:29:28,593][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:29:29,093][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:29:29,593][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:29:30,093][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:29:30,594][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:29:31,093][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:29:31,593][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:29:32,093][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:29:32,590][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:29:33,087][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:29:33,585][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:29:34,082][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:29:34,579][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:29:35,077][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:29:35,576][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:29:36,075][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:29:36,576][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:29:37,075][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:29:37,575][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:29:38,074][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:29:38,573][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:29:39,075][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:29:39,574][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:29:40,073][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:29:40,574][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:29:41,074][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:29:41,574][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:29:42,075][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:29:42,577][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:29:43,078][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:29:43,579][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:29:44,081][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:29:44,580][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:29:45,080][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:29:45,582][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:29:46,084][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:29:46,583][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:29:47,083][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:29:47,582][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:29:48,083][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:29:48,584][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:29:49,083][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:29:49,581][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:29:50,080][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:29:50,579][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:29:51,078][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:29:51,577][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:29:52,076][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:29:52,575][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:29:53,074][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:29:53,573][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:29:54,073][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:29:54,572][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:29:55,071][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:29:55,569][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:29:56,068][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:29:56,566][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:29:57,065][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:29:57,564][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:29:58,064][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:29:58,564][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:29:59,063][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:29:59,562][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:30:00,061][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:30:00,561][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:30:01,060][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:30:01,559][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:30:02,060][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:30:02,560][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:30:03,059][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:30:03,559][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:30:04,059][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:30:04,557][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:30:05,054][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:30:05,552][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:30:06,051][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:30:06,547][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:30:07,044][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:30:07,541][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:30:08,039][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:30:08,672][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.14%, Current % of VRAM taken: 60.50%, Block Peak % of device VRAM: 62.35%, ΔTime: 00:01:04 [2026-03-26 08:30:09,414][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:30:09,416][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:30:09,418][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:30:10,074][__main__][INFO] - Iteration 566 took 1m 14s (9.44% Gen, 89.68% Train). Generation: 7s, Training: 1m 6s. Estimated remaining time: 50h 9m 30s. Estimated total time: 62h 4m 58s. Time estimates for 10 more iterations: 12m 24s, 100 more iterations: 2h 4m 9s, 500 more iterations: 10h 20m 49s. [2026-03-26 08:30:10,076][__main__][INFO] - Starting iteration 566. [2026-03-26 08:30:10,476][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:30:10,477][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:30:17,163][__main__][INFO] - Number of regex retries in iteration 566: 0 [2026-03-26 08:30:17,164][__main__][INFO] - agents played in iteration 566 are Alice, Bob [2026-03-26 08:30:18,044][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:30:18,090][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.45%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:30:18,091][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:30:18,091][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:30:18,696][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:30:19,152][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:30:19,656][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:30:20,156][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:30:20,656][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:30:21,155][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:30:21,655][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:30:22,154][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:30:22,653][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:30:23,154][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:30:23,655][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:30:24,155][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:30:24,655][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:30:25,155][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:30:25,655][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:30:26,155][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:30:26,654][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:30:27,153][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:30:27,653][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:30:28,152][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:30:28,652][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:30:29,150][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:30:29,649][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:30:30,147][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:30:30,645][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:30:31,145][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:30:31,644][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:30:32,142][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:30:32,642][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:30:33,141][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:30:33,641][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:30:34,141][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:30:34,640][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:30:35,139][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:30:35,638][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:30:36,136][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:30:36,635][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:30:37,133][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:30:37,632][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:30:38,131][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:30:38,629][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:30:39,129][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:30:39,627][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:30:40,127][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:30:40,626][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:30:41,125][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:30:41,624][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:30:42,123][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:30:42,622][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:30:43,122][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:30:43,621][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:30:44,121][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:30:44,621][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:30:45,121][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:30:45,621][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:30:46,121][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:30:46,620][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:30:47,122][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:30:47,623][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:30:48,123][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:30:48,625][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:30:49,125][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:30:49,625][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:30:50,125][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:30:50,626][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:30:51,126][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:30:51,626][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:30:52,126][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:30:52,625][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:30:53,125][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:30:53,625][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:30:54,123][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:30:54,622][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:30:55,122][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:30:55,622][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:30:56,121][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:30:56,620][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:30:57,120][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:30:57,620][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:30:58,120][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:30:58,620][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:30:59,118][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:30:59,617][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:31:00,115][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:31:00,612][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:31:01,110][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:31:01,607][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:31:02,105][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:31:02,602][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:31:03,102][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:31:03,602][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:31:04,101][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:31:04,601][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:31:05,101][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:31:05,601][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:31:06,101][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:31:06,601][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:31:07,100][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:31:07,599][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:31:08,098][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:31:08,597][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:31:09,097][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:31:09,596][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:31:10,095][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:31:10,593][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:31:11,092][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:31:11,591][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:31:12,090][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:31:12,589][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:31:13,087][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:31:13,586][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:31:14,084][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:31:14,582][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:31:15,082][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:31:15,585][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:31:16,087][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:31:16,586][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:31:17,085][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:31:17,584][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:31:18,084][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:31:18,586][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:31:19,086][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:31:19,587][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:31:20,086][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:31:20,586][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:31:21,087][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:31:21,587][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:31:22,087][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:31:22,587][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21758 tokens. [2026-03-26 08:31:23,224][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.30%, Current % of VRAM taken: 60.65%, Block Peak % of device VRAM: 62.41%, ΔTime: 00:01:04 [2026-03-26 08:31:23,964][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:31:23,966][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:31:23,968][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:31:24,633][__main__][INFO] - Iteration 567 took 1m 14s (9.02% Gen, 90.08% Train). Generation: 6s, Training: 1m 6s. Estimated remaining time: 49h 51m 10s. Estimated total time: 61h 47m 53s. Time estimates for 10 more iterations: 12m 21s, 100 more iterations: 2h 3m 35s, 500 more iterations: 10h 17m 58s. [2026-03-26 08:31:24,636][__main__][INFO] - Starting iteration 567. [2026-03-26 08:31:25,039][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:31:25,040][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:31:25,639][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:31:27,281][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:31:31,680][__main__][INFO] - Number of regex retries in iteration 567: 2 [2026-03-26 08:31:31,680][__main__][INFO] - agents played in iteration 567 are Alice, Bob [2026-03-26 08:31:32,560][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:31:32,606][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.51%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:31:32,606][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:31:32,607][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:31:33,481][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:31:33,936][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:31:34,440][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:31:34,938][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:31:35,437][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:31:35,935][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:31:36,434][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:31:36,932][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:31:37,431][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:31:37,933][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:31:38,433][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:31:38,933][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:31:39,433][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:31:39,933][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:31:40,432][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:31:40,933][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:31:41,433][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:31:41,930][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:31:42,428][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:31:42,926][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:31:43,423][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:31:43,921][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:31:44,419][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:31:44,918][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:31:45,416][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:31:45,916][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:31:46,416][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:31:46,915][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:31:47,415][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:31:47,915][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:31:48,416][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:31:48,915][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:31:49,415][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:31:49,914][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:31:50,413][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:31:50,913][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:31:51,412][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:31:51,912][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:31:52,411][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:31:52,910][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:31:53,408][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:31:53,906][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:31:54,404][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:31:54,902][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:31:55,401][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:31:55,899][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:31:56,398][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:31:56,897][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:31:57,395][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:31:57,893][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:31:58,392][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:31:58,891][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:31:59,390][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:31:59,889][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:32:00,387][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:32:00,885][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:32:01,384][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:32:01,882][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:32:02,381][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:32:02,881][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:32:03,381][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:32:03,881][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:32:04,380][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:32:04,879][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:32:05,378][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:32:05,878][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:32:06,377][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:32:06,876][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:32:07,375][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:32:07,873][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:32:08,372][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:32:08,872][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:32:09,372][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:32:09,872][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:32:10,373][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:32:10,873][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:32:11,374][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:32:11,874][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:32:12,376][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:32:12,876][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:32:13,376][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:32:13,875][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:32:14,373][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:32:14,871][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:32:15,369][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:32:15,867][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:32:16,365][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:32:16,863][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:32:17,361][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:32:17,861][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:32:18,363][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:32:18,863][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:32:19,362][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:32:19,860][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:32:20,359][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:32:20,859][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:32:21,358][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:32:21,859][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:32:22,358][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:32:22,857][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:32:23,357][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:32:23,857][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:32:24,357][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:32:24,856][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:32:25,358][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:32:25,857][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:32:26,355][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:32:26,854][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:32:27,354][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:32:27,852][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:32:28,353][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:32:28,852][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:32:29,351][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:32:29,850][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:32:30,348][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:32:30,847][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:32:31,346][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:32:31,845][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:32:32,344][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:32:32,844][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:32:33,344][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:32:33,843][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:32:34,341][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:32:34,840][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:32:35,339][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:32:35,838][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:32:36,337][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:32:36,837][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:32:37,336][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21759 tokens. [2026-03-26 08:32:37,973][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.20%, Current % of VRAM taken: 60.55%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:04 [2026-03-26 08:32:38,712][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:32:38,715][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:32:38,716][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:32:39,366][__main__][INFO] - Iteration 568 took 1m 14s (8.93% Gen, 90.19% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 49h 58m 23s. Estimated total time: 61h 56m 20s. Time estimates for 10 more iterations: 12m 23s, 100 more iterations: 2h 3m 52s, 500 more iterations: 10h 19m 23s. [2026-03-26 08:32:39,368][__main__][INFO] - Starting iteration 568. [2026-03-26 08:32:39,771][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:32:39,772][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:32:41,996][mllm.models.large_language_model_local][WARNING] - Response Proposal: 10 hats, 10 books, 10 balls did not match regex: (?i)Proposal:\s*((?:\s*(?P(10|[0-9]))\s*(?Phats?|books?|balls?)\s*,?)+), retry 1/3 [2026-03-26 08:32:46,567][__main__][INFO] - Number of regex retries in iteration 568: 1 [2026-03-26 08:32:46,568][__main__][INFO] - agents played in iteration 568 are Alice, Bob [2026-03-26 08:32:47,446][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:32:47,491][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.46%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:32:47,491][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:32:47,492][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:32:48,101][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:32:48,556][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:32:49,060][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:32:49,560][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:32:50,059][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:32:50,558][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:32:51,057][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:32:51,555][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:32:52,055][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:32:52,556][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:32:53,056][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:32:53,556][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:32:54,056][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:32:54,555][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:32:55,054][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:32:55,553][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:32:56,054][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:32:56,553][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:32:57,052][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:32:57,550][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:32:58,050][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:32:58,548][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:32:59,046][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:32:59,545][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:33:00,043][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:33:00,540][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:33:01,039][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:33:01,538][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:33:02,038][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:33:02,536][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:33:03,035][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:33:03,534][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:33:04,034][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:33:04,533][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:33:05,032][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:33:05,531][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:33:06,030][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:33:06,528][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:33:07,026][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:33:07,526][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:33:08,025][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:33:08,527][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:33:09,030][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:33:09,531][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:33:10,031][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:33:10,922][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:33:11,523][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:33:12,023][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128 [2026-03-26 08:33:12,522][mllm.training.trainer_common][INFO] - Processing mini-batch 48 of 128 [2026-03-26 08:33:13,022][mllm.training.trainer_common][INFO] - Processing mini-batch 49 of 128 [2026-03-26 08:33:13,523][mllm.training.trainer_common][INFO] - Processing mini-batch 50 of 128 [2026-03-26 08:33:14,022][mllm.training.trainer_common][INFO] - Processing mini-batch 51 of 128 [2026-03-26 08:33:14,522][mllm.training.trainer_common][INFO] - Processing mini-batch 52 of 128 [2026-03-26 08:33:15,021][mllm.training.trainer_common][INFO] - Processing mini-batch 53 of 128 [2026-03-26 08:33:15,521][mllm.training.trainer_common][INFO] - Processing mini-batch 54 of 128 [2026-03-26 08:33:16,021][mllm.training.trainer_common][INFO] - Processing mini-batch 55 of 128 [2026-03-26 08:33:16,521][mllm.training.trainer_common][INFO] - Processing mini-batch 56 of 128 [2026-03-26 08:33:17,021][mllm.training.trainer_common][INFO] - Processing mini-batch 57 of 128 [2026-03-26 08:33:17,520][mllm.training.trainer_common][INFO] - Processing mini-batch 58 of 128 [2026-03-26 08:33:18,020][mllm.training.trainer_common][INFO] - Processing mini-batch 59 of 128 [2026-03-26 08:33:18,522][mllm.training.trainer_common][INFO] - Processing mini-batch 60 of 128 [2026-03-26 08:33:19,021][mllm.training.trainer_common][INFO] - Processing mini-batch 61 of 128 [2026-03-26 08:33:19,521][mllm.training.trainer_common][INFO] - Processing mini-batch 62 of 128 [2026-03-26 08:33:20,020][mllm.training.trainer_common][INFO] - Processing mini-batch 63 of 128 [2026-03-26 08:33:20,519][mllm.training.trainer_common][INFO] - Processing mini-batch 64 of 128 [2026-03-26 08:33:21,017][mllm.training.trainer_common][INFO] - Processing mini-batch 65 of 128 [2026-03-26 08:33:21,516][mllm.training.trainer_common][INFO] - Processing mini-batch 66 of 128 [2026-03-26 08:33:22,014][mllm.training.trainer_common][INFO] - Processing mini-batch 67 of 128 [2026-03-26 08:33:22,512][mllm.training.trainer_common][INFO] - Processing mini-batch 68 of 128 [2026-03-26 08:33:23,010][mllm.training.trainer_common][INFO] - Processing mini-batch 69 of 128 [2026-03-26 08:33:23,508][mllm.training.trainer_common][INFO] - Processing mini-batch 70 of 128 [2026-03-26 08:33:24,006][mllm.training.trainer_common][INFO] - Processing mini-batch 71 of 128 [2026-03-26 08:33:24,504][mllm.training.trainer_common][INFO] - Processing mini-batch 72 of 128 [2026-03-26 08:33:25,004][mllm.training.trainer_common][INFO] - Processing mini-batch 73 of 128 [2026-03-26 08:33:25,504][mllm.training.trainer_common][INFO] - Processing mini-batch 74 of 128 [2026-03-26 08:33:26,003][mllm.training.trainer_common][INFO] - Processing mini-batch 75 of 128 [2026-03-26 08:33:26,503][mllm.training.trainer_common][INFO] - Processing mini-batch 76 of 128 [2026-03-26 08:33:27,003][mllm.training.trainer_common][INFO] - Processing mini-batch 77 of 128 [2026-03-26 08:33:27,502][mllm.training.trainer_common][INFO] - Processing mini-batch 78 of 128 [2026-03-26 08:33:28,002][mllm.training.trainer_common][INFO] - Processing mini-batch 79 of 128 [2026-03-26 08:33:28,502][mllm.training.trainer_common][INFO] - Processing mini-batch 80 of 128 [2026-03-26 08:33:29,000][mllm.training.trainer_common][INFO] - Processing mini-batch 81 of 128 [2026-03-26 08:33:29,499][mllm.training.trainer_common][INFO] - Processing mini-batch 82 of 128 [2026-03-26 08:33:29,997][mllm.training.trainer_common][INFO] - Processing mini-batch 83 of 128 [2026-03-26 08:33:30,495][mllm.training.trainer_common][INFO] - Processing mini-batch 84 of 128 [2026-03-26 08:33:30,993][mllm.training.trainer_common][INFO] - Processing mini-batch 85 of 128 [2026-03-26 08:33:31,492][mllm.training.trainer_common][INFO] - Processing mini-batch 86 of 128 [2026-03-26 08:33:31,991][mllm.training.trainer_common][INFO] - Processing mini-batch 87 of 128 [2026-03-26 08:33:32,489][mllm.training.trainer_common][INFO] - Processing mini-batch 88 of 128 [2026-03-26 08:33:32,988][mllm.training.trainer_common][INFO] - Processing mini-batch 89 of 128 [2026-03-26 08:33:33,487][mllm.training.trainer_common][INFO] - Processing mini-batch 90 of 128 [2026-03-26 08:33:33,986][mllm.training.trainer_common][INFO] - Processing mini-batch 91 of 128 [2026-03-26 08:33:34,484][mllm.training.trainer_common][INFO] - Processing mini-batch 92 of 128 [2026-03-26 08:33:34,983][mllm.training.trainer_common][INFO] - Processing mini-batch 93 of 128 [2026-03-26 08:33:35,481][mllm.training.trainer_common][INFO] - Processing mini-batch 94 of 128 [2026-03-26 08:33:35,981][mllm.training.trainer_common][INFO] - Processing mini-batch 95 of 128 [2026-03-26 08:33:36,480][mllm.training.trainer_common][INFO] - Processing mini-batch 96 of 128 [2026-03-26 08:33:36,979][mllm.training.trainer_common][INFO] - Processing mini-batch 97 of 128 [2026-03-26 08:33:37,478][mllm.training.trainer_common][INFO] - Processing mini-batch 98 of 128 [2026-03-26 08:33:37,978][mllm.training.trainer_common][INFO] - Processing mini-batch 99 of 128 [2026-03-26 08:33:38,476][mllm.training.trainer_common][INFO] - Processing mini-batch 100 of 128 [2026-03-26 08:33:38,976][mllm.training.trainer_common][INFO] - Processing mini-batch 101 of 128 [2026-03-26 08:33:39,476][mllm.training.trainer_common][INFO] - Processing mini-batch 102 of 128 [2026-03-26 08:33:39,975][mllm.training.trainer_common][INFO] - Processing mini-batch 103 of 128 [2026-03-26 08:33:40,473][mllm.training.trainer_common][INFO] - Processing mini-batch 104 of 128 [2026-03-26 08:33:40,974][mllm.training.trainer_common][INFO] - Processing mini-batch 105 of 128 [2026-03-26 08:33:41,474][mllm.training.trainer_common][INFO] - Processing mini-batch 106 of 128 [2026-03-26 08:33:41,975][mllm.training.trainer_common][INFO] - Processing mini-batch 107 of 128 [2026-03-26 08:33:42,475][mllm.training.trainer_common][INFO] - Processing mini-batch 108 of 128 [2026-03-26 08:33:42,976][mllm.training.trainer_common][INFO] - Processing mini-batch 109 of 128 [2026-03-26 08:33:43,476][mllm.training.trainer_common][INFO] - Processing mini-batch 110 of 128 [2026-03-26 08:33:43,977][mllm.training.trainer_common][INFO] - Processing mini-batch 111 of 128 [2026-03-26 08:33:44,479][mllm.training.trainer_common][INFO] - Processing mini-batch 112 of 128 [2026-03-26 08:33:44,979][mllm.training.trainer_common][INFO] - Processing mini-batch 113 of 128 [2026-03-26 08:33:45,480][mllm.training.trainer_common][INFO] - Processing mini-batch 114 of 128 [2026-03-26 08:33:45,981][mllm.training.trainer_common][INFO] - Processing mini-batch 115 of 128 [2026-03-26 08:33:46,482][mllm.training.trainer_common][INFO] - Processing mini-batch 116 of 128 [2026-03-26 08:33:46,983][mllm.training.trainer_common][INFO] - Processing mini-batch 117 of 128 [2026-03-26 08:33:47,482][mllm.training.trainer_common][INFO] - Processing mini-batch 118 of 128 [2026-03-26 08:33:47,983][mllm.training.trainer_common][INFO] - Processing mini-batch 119 of 128 [2026-03-26 08:33:48,485][mllm.training.trainer_common][INFO] - Processing mini-batch 120 of 128 [2026-03-26 08:33:48,984][mllm.training.trainer_common][INFO] - Processing mini-batch 121 of 128 [2026-03-26 08:33:49,483][mllm.training.trainer_common][INFO] - Processing mini-batch 122 of 128 [2026-03-26 08:33:49,984][mllm.training.trainer_common][INFO] - Processing mini-batch 123 of 128 [2026-03-26 08:33:50,483][mllm.training.trainer_common][INFO] - Processing mini-batch 124 of 128 [2026-03-26 08:33:50,983][mllm.training.trainer_common][INFO] - Processing mini-batch 125 of 128 [2026-03-26 08:33:51,483][mllm.training.trainer_common][INFO] - Processing mini-batch 126 of 128 [2026-03-26 08:33:51,982][mllm.training.trainer_common][INFO] - Processing mini-batch 127 of 128 [2026-03-26 08:33:52,484][mllm.training.trainer_common][INFO] - Accumulated the policy gradient loss for 21757 tokens. [2026-03-26 08:33:53,130][mllm.training.trainer_common][INFO] - For task: Apply reinforce step, ΔVRAM % (total): 14.23%, Current % of VRAM taken: 60.58%, Block Peak % of device VRAM: 62.31%, ΔTime: 00:01:05 [2026-03-26 08:33:53,906][mllm.training.trainer_common][INFO] - Saved main optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/policy_optimizer_state.pt [2026-03-26 08:33:53,908][mllm.training.trainer_common][INFO] - Saved critic optimizer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/critic_optimizer_state.pt [2026-03-26 08:33:53,910][mllm.training.trainer_common][INFO] - Saved trainer state to /scratch/muqeeth/llm_negotiation/2026_03/split_no_comm_vanilla_ad_align_no_agent_buffer_seed1337/seed_1337/agent_trainer/trainer_annealing_state.pkl [2026-03-26 08:33:54,826][__main__][INFO] - Iteration 569 took 1m 15s (9.05% Gen, 89.72% Train). Generation: 6s, Training: 1m 7s. Estimated remaining time: 50h 33m 35s. Estimated total time: 62h 32m 48s. Time estimates for 10 more iterations: 12m 30s, 100 more iterations: 2h 5m 5s, 500 more iterations: 10h 25m 28s. [2026-03-26 08:33:54,831][__main__][INFO] - Starting iteration 569. [2026-03-26 08:33:55,231][__main__][INFO] - Inference policies count is regular policies 2 and buffer policies 11 and human policies 1. [2026-03-26 08:33:55,232][__main__][INFO] - Hard coded buffer agents are set to False with prob 0 [2026-03-26 08:34:02,515][__main__][INFO] - Number of regex retries in iteration 569: 0 [2026-03-26 08:34:02,516][__main__][INFO] - agents played in iteration 569 are Alice, Bob [2026-03-26 08:34:03,528][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:34:03,573][mllm.training.trainer_ad_align][INFO] - For task: Get advantages with critic gradient accumulation, ΔVRAM % (total): 0.00%, Current % of VRAM taken: 53.47%, Block Peak % of device VRAM: 19.53%, ΔTime: 00:00:00 [2026-03-26 08:34:03,574][mllm.training.trainer_ad_align][INFO] - Sharing advantage alignment data. [2026-03-26 08:34:03,574][mllm.training.trainer_ad_align][INFO] - Receiving advantage packets. [2026-03-26 08:34:04,289][mllm.training.trainer_common][INFO] - Processing mini-batch 0 of 128 [2026-03-26 08:34:04,745][mllm.training.trainer_common][INFO] - Processing mini-batch 1 of 128 [2026-03-26 08:34:05,248][mllm.training.trainer_common][INFO] - Processing mini-batch 2 of 128 [2026-03-26 08:34:05,747][mllm.training.trainer_common][INFO] - Processing mini-batch 3 of 128 [2026-03-26 08:34:06,245][mllm.training.trainer_common][INFO] - Processing mini-batch 4 of 128 [2026-03-26 08:34:06,743][mllm.training.trainer_common][INFO] - Processing mini-batch 5 of 128 [2026-03-26 08:34:07,242][mllm.training.trainer_common][INFO] - Processing mini-batch 6 of 128 [2026-03-26 08:34:07,740][mllm.training.trainer_common][INFO] - Processing mini-batch 7 of 128 [2026-03-26 08:34:08,239][mllm.training.trainer_common][INFO] - Processing mini-batch 8 of 128 [2026-03-26 08:34:08,740][mllm.training.trainer_common][INFO] - Processing mini-batch 9 of 128 [2026-03-26 08:34:09,240][mllm.training.trainer_common][INFO] - Processing mini-batch 10 of 128 [2026-03-26 08:34:09,740][mllm.training.trainer_common][INFO] - Processing mini-batch 11 of 128 [2026-03-26 08:34:10,239][mllm.training.trainer_common][INFO] - Processing mini-batch 12 of 128 [2026-03-26 08:34:10,738][mllm.training.trainer_common][INFO] - Processing mini-batch 13 of 128 [2026-03-26 08:34:11,237][mllm.training.trainer_common][INFO] - Processing mini-batch 14 of 128 [2026-03-26 08:34:11,736][mllm.training.trainer_common][INFO] - Processing mini-batch 15 of 128 [2026-03-26 08:34:12,236][mllm.training.trainer_common][INFO] - Processing mini-batch 16 of 128 [2026-03-26 08:34:12,738][mllm.training.trainer_common][INFO] - Processing mini-batch 17 of 128 [2026-03-26 08:34:13,237][mllm.training.trainer_common][INFO] - Processing mini-batch 18 of 128 [2026-03-26 08:34:13,736][mllm.training.trainer_common][INFO] - Processing mini-batch 19 of 128 [2026-03-26 08:34:14,236][mllm.training.trainer_common][INFO] - Processing mini-batch 20 of 128 [2026-03-26 08:34:14,735][mllm.training.trainer_common][INFO] - Processing mini-batch 21 of 128 [2026-03-26 08:34:15,236][mllm.training.trainer_common][INFO] - Processing mini-batch 22 of 128 [2026-03-26 08:34:15,739][mllm.training.trainer_common][INFO] - Processing mini-batch 23 of 128 [2026-03-26 08:34:16,242][mllm.training.trainer_common][INFO] - Processing mini-batch 24 of 128 [2026-03-26 08:34:16,742][mllm.training.trainer_common][INFO] - Processing mini-batch 25 of 128 [2026-03-26 08:34:17,242][mllm.training.trainer_common][INFO] - Processing mini-batch 26 of 128 [2026-03-26 08:34:17,742][mllm.training.trainer_common][INFO] - Processing mini-batch 27 of 128 [2026-03-26 08:34:18,244][mllm.training.trainer_common][INFO] - Processing mini-batch 28 of 128 [2026-03-26 08:34:18,744][mllm.training.trainer_common][INFO] - Processing mini-batch 29 of 128 [2026-03-26 08:34:19,244][mllm.training.trainer_common][INFO] - Processing mini-batch 30 of 128 [2026-03-26 08:34:19,745][mllm.training.trainer_common][INFO] - Processing mini-batch 31 of 128 [2026-03-26 08:34:20,245][mllm.training.trainer_common][INFO] - Processing mini-batch 32 of 128 [2026-03-26 08:34:20,743][mllm.training.trainer_common][INFO] - Processing mini-batch 33 of 128 [2026-03-26 08:34:21,241][mllm.training.trainer_common][INFO] - Processing mini-batch 34 of 128 [2026-03-26 08:34:21,739][mllm.training.trainer_common][INFO] - Processing mini-batch 35 of 128 [2026-03-26 08:34:22,238][mllm.training.trainer_common][INFO] - Processing mini-batch 36 of 128 [2026-03-26 08:34:22,736][mllm.training.trainer_common][INFO] - Processing mini-batch 37 of 128 [2026-03-26 08:34:23,234][mllm.training.trainer_common][INFO] - Processing mini-batch 38 of 128 [2026-03-26 08:34:23,733][mllm.training.trainer_common][INFO] - Processing mini-batch 39 of 128 [2026-03-26 08:34:24,232][mllm.training.trainer_common][INFO] - Processing mini-batch 40 of 128 [2026-03-26 08:34:24,730][mllm.training.trainer_common][INFO] - Processing mini-batch 41 of 128 [2026-03-26 08:34:25,228][mllm.training.trainer_common][INFO] - Processing mini-batch 42 of 128 [2026-03-26 08:34:25,728][mllm.training.trainer_common][INFO] - Processing mini-batch 43 of 128 [2026-03-26 08:34:26,226][mllm.training.trainer_common][INFO] - Processing mini-batch 44 of 128 [2026-03-26 08:34:26,724][mllm.training.trainer_common][INFO] - Processing mini-batch 45 of 128 [2026-03-26 08:34:27,222][mllm.training.trainer_common][INFO] - Processing mini-batch 46 of 128 [2026-03-26 08:34:27,722][mllm.training.trainer_common][INFO] - Processing mini-batch 47 of 128