2026-08-10T13:47:58-07:00 serving outputs/release/healed/release-offpolicy-math-keep50/step0100 on GPU 1 port 8491 (pp=1 think=template-default) 2026-08-10T13:47:58-07:00 waiting for server /health ... 2026-08-10T13:48:48-07:00 server up; chat pass [gsm8k_cot_zeroshot,gsm8k_cot,minerva_math500,ifeval] 2026-08-10:13:48:55 INFO [_cli.run:388] Selected Tasks: ['gsm8k_cot_zeroshot', 'gsm8k_cot', 'minerva_math500', 'ifeval'] 2026-08-10:13:48:56 INFO [evaluator:214] Setting random seed to 0 | Setting numpy seed to 1234 | Setting torch manual seed to 1234 | Setting fewshot manual seed to 1234 2026-08-10:13:48:56 WARNING [evaluator:226] generation_kwargs: {'max_gen_toks': 1280} specified through cli, these settings will update set parameters in yaml tasks. Ensure 'do_sample=True' for non-greedy decoding! 2026-08-10:13:48:56 INFO [evaluator:239] Initializing local-chat-completions model, with arguments: {'model': 'student', 'base_url': 'http://127.0.0.1:8491/v1/chat/completions', 'num_concurrent': 48, 'tokenized_requests': False, 'max_retries': 3} 2026-08-10:13:48:56 INFO [models.api_models:179] Using max length 2048 - 1 2026-08-10:13:48:56 INFO [models.api_models:200] Using tokenizer None 2026-08-10:13:49:02 INFO [evaluator_utils:446] Selected tasks: 2026-08-10:13:49:02 INFO [evaluator_utils:480] Task: gsm8k_cot (gsm8k/gsm8k-cot.yaml) 2026-08-10:13:49:02 INFO [evaluator_utils:480] Task: gsm8k_cot_zeroshot (gsm8k/gsm8k-cot-zeroshot.yaml) 2026-08-10:13:49:02 INFO [evaluator_utils:480] Task: ifeval (ifeval/ifeval.yaml) 2026-08-10:13:49:02 INFO [evaluator_utils:480] Task: minerva_math500 (minerva_math/minerva_math500.yaml) 2026-08-10:13:49:02 INFO [evaluator:314] gsm8k_cot_zeroshot: Using gen_kwargs: {'until': ['Q:', '', '<|im_end|>'], 'do_sample': False, 'max_gen_toks': 1280} 2026-08-10:13:49:02 INFO [evaluator:314] gsm8k_cot: Using gen_kwargs: {'do_sample': False, 'until': ['Q:', '', '<|im_end|>'], 'max_gen_toks': 1280} 2026-08-10:13:49:02 INFO [evaluator:314] minerva_math500: Using gen_kwargs: {'until': ['Problem:'], 'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 1280} 2026-08-10:13:49:02 INFO [evaluator:314] ifeval: Using gen_kwargs: {'until': [], 'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 1280} 2026-08-10:13:49:02 INFO [api.task:312] Building contexts for gsm8k_cot_zeroshot on rank 0... 0%| | 0/1319 [00:00 outputs/evals/redo_policy/offpolicy_step0100