| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9996190476190476, | |
| "eval_steps": 50, | |
| "global_step": 164, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.23935546875, | |
| "completions/max_length": 50.6, | |
| "completions/max_terminated_length": 50.4, | |
| "completions/mean_length": 48.8724609375, | |
| "completions/mean_terminated_length": 48.6720588684082, | |
| "completions/min_length": 2.0, | |
| "completions/min_terminated_length": 2.0, | |
| "epoch": 0.030476190476190476, | |
| "grad_norm": 0.9191175103187561, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.014, | |
| "num_tokens": 1811174.0, | |
| "reward": 0.10400769710540772, | |
| "reward_std": 0.5274705350399017, | |
| "rewards/accuracy_reward": 0.00712890625, | |
| "rewards/brier_reward": 0.007832830213010311, | |
| "rewards/confidence_one_or_zero": 0.0103515625, | |
| "rewards/format_reward": 0.01064453125, | |
| "rewards/log_2_reward": 0.09512097388505936, | |
| "rewards/mean_confidence_reward": 0.035655312985181806, | |
| "step": 5 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.3904296875, | |
| "completions/max_length": 50.6, | |
| "completions/max_terminated_length": 50.4, | |
| "completions/mean_length": 48.72880859375, | |
| "completions/mean_terminated_length": 48.19507522583008, | |
| "completions/min_length": 2.2, | |
| "completions/min_terminated_length": 2.2, | |
| "epoch": 0.06095238095238095, | |
| "grad_norm": 0.16941173374652863, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0316, | |
| "num_tokens": 3620877.0, | |
| "reward": 0.2629227817058563, | |
| "reward_std": 1.1119458675384521, | |
| "rewards/accuracy_reward": 0.0177734375, | |
| "rewards/brier_reward": 0.019864578917622566, | |
| "rewards/confidence_one_or_zero": 0.0130859375, | |
| "rewards/format_reward": 0.02734375, | |
| "rewards/log_2_reward": 0.2403641790151596, | |
| "rewards/mean_confidence_reward": 0.05027429163455963, | |
| "step": 10 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.1251953125, | |
| "completions/max_length": 50.2, | |
| "completions/max_terminated_length": 50.2, | |
| "completions/mean_length": 45.181640625, | |
| "completions/mean_terminated_length": 44.80078964233398, | |
| "completions/min_length": 8.6, | |
| "completions/min_terminated_length": 8.6, | |
| "epoch": 0.09142857142857143, | |
| "grad_norm": 1.1244808435440063, | |
| "learning_rate": 1e-06, | |
| "loss": 0.245, | |
| "num_tokens": 5394257.0, | |
| "reward": 4.930716705322266, | |
| "reward_std": 4.177352714538574, | |
| "rewards/accuracy_reward": 0.35673828125, | |
| "rewards/brier_reward": 0.37847294211387633, | |
| "rewards/confidence_one_or_zero": 0.07822265625, | |
| "rewards/format_reward": 0.50244140625, | |
| "rewards/log_2_reward": 4.501126909255982, | |
| "rewards/mean_confidence_reward": 0.47802442610263823, | |
| "step": 15 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0080078125, | |
| "completions/max_length": 49.8, | |
| "completions/max_terminated_length": 49.8, | |
| "completions/mean_length": 38.0888671875, | |
| "completions/mean_terminated_length": 38.002176666259764, | |
| "completions/min_length": 17.0, | |
| "completions/min_terminated_length": 17.0, | |
| "epoch": 0.1219047619047619, | |
| "grad_norm": 0.04089135304093361, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0241, | |
| "num_tokens": 7095007.0, | |
| "reward": 9.922300148010255, | |
| "reward_std": 1.1487209796905518, | |
| "rewards/accuracy_reward": 0.74013671875, | |
| "rewards/brier_reward": 0.7714210748672485, | |
| "rewards/confidence_one_or_zero": 0.0666015625, | |
| "rewards/format_reward": 0.98447265625, | |
| "rewards/log_2_reward": 9.059995079040528, | |
| "rewards/mean_confidence_reward": 0.9122045636177063, | |
| "step": 20 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00048828125, | |
| "completions/max_length": 45.4, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 36.56953125, | |
| "completions/mean_terminated_length": 36.56360931396485, | |
| "completions/min_length": 26.8, | |
| "completions/min_terminated_length": 26.8, | |
| "epoch": 0.1523809523809524, | |
| "grad_norm": 0.0074047367088496685, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0001, | |
| "num_tokens": 8780199.0, | |
| "reward": 10.27893524169922, | |
| "reward_std": 0.22245991080999375, | |
| "rewards/accuracy_reward": 0.75419921875, | |
| "rewards/brier_reward": 0.8098533034324646, | |
| "rewards/confidence_one_or_zero": 0.00341796875, | |
| "rewards/format_reward": 0.99912109375, | |
| "rewards/log_2_reward": 9.402275276184081, | |
| "rewards/mean_confidence_reward": 0.8197431087493896, | |
| "step": 25 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 9.765625e-05, | |
| "completions/max_length": 43.0, | |
| "completions/max_terminated_length": 41.4, | |
| "completions/mean_length": 34.4681640625, | |
| "completions/mean_terminated_length": 34.46673583984375, | |
| "completions/min_length": 32.0, | |
| "completions/min_terminated_length": 32.0, | |
| "epoch": 0.18285714285714286, | |
| "grad_norm": 0.010611865669488907, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0004, | |
| "num_tokens": 10443873.0, | |
| "reward": 10.285838317871093, | |
| "reward_std": 0.1147611990571022, | |
| "rewards/accuracy_reward": 0.73349609375, | |
| "rewards/brier_reward": 0.8037224054336548, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 0.9998046875, | |
| "rewards/log_2_reward": 9.419188308715821, | |
| "rewards/mean_confidence_reward": 0.6957599520683289, | |
| "step": 30 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 43.2, | |
| "completions/max_terminated_length": 43.2, | |
| "completions/mean_length": 35.34462890625, | |
| "completions/mean_terminated_length": 35.34462890625, | |
| "completions/min_length": 32.4, | |
| "completions/min_terminated_length": 32.4, | |
| "epoch": 0.21333333333333335, | |
| "grad_norm": 0.004959334619343281, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 12116522.0, | |
| "reward": 10.252646255493165, | |
| "reward_std": 0.06293957829475402, | |
| "rewards/accuracy_reward": 0.71875, | |
| "rewards/brier_reward": 0.7925641775131226, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.393271446228027, | |
| "rewards/mean_confidence_reward": 0.6929210901260376, | |
| "step": 35 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.6, | |
| "completions/max_terminated_length": 41.6, | |
| "completions/mean_length": 36.130859375, | |
| "completions/mean_terminated_length": 36.130859375, | |
| "completions/min_length": 32.2, | |
| "completions/min_terminated_length": 32.2, | |
| "epoch": 0.2438095238095238, | |
| "grad_norm": 0.004302784334868193, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 13797222.0, | |
| "reward": 10.311922454833985, | |
| "reward_std": 0.043248052895069125, | |
| "rewards/accuracy_reward": 0.746875, | |
| "rewards/brier_reward": 0.8125308394432068, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.438485145568848, | |
| "rewards/mean_confidence_reward": 0.6959580302238464, | |
| "step": 40 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 38.8, | |
| "completions/max_terminated_length": 38.8, | |
| "completions/mean_length": 37.3640625, | |
| "completions/mean_terminated_length": 37.3640625, | |
| "completions/min_length": 32.8, | |
| "completions/min_terminated_length": 32.8, | |
| "epoch": 0.2742857142857143, | |
| "grad_norm": 0.0034340149722993374, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 15490550.0, | |
| "reward": 10.362943267822265, | |
| "reward_std": 0.030235714837908746, | |
| "rewards/accuracy_reward": 0.78740234375, | |
| "rewards/brier_reward": 0.8271316170692444, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.469242095947266, | |
| "rewards/mean_confidence_reward": 0.6924892663955688, | |
| "step": 45 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 9.765625e-05, | |
| "completions/max_length": 42.8, | |
| "completions/max_terminated_length": 41.2, | |
| "completions/mean_length": 37.9048828125, | |
| "completions/mean_terminated_length": 37.90370407104492, | |
| "completions/min_length": 33.0, | |
| "completions/min_terminated_length": 33.0, | |
| "epoch": 0.3047619047619048, | |
| "grad_norm": 0.0021895961835980415, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0002, | |
| "num_tokens": 17189416.0, | |
| "reward": 10.276718902587891, | |
| "reward_std": 0.03380730748176575, | |
| "rewards/accuracy_reward": 0.721875, | |
| "rewards/brier_reward": 0.8022076249122619, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 0.99990234375, | |
| "rewards/log_2_reward": 9.415830421447755, | |
| "rewards/mean_confidence_reward": 0.7136552929878235, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.3047619047619048, | |
| "eval_completions/clipped_ratio": 0.0, | |
| "eval_completions/max_length": 40.0, | |
| "eval_completions/max_terminated_length": 40.0, | |
| "eval_completions/mean_length": 37.991783142089844, | |
| "eval_completions/mean_terminated_length": 37.991783142089844, | |
| "eval_completions/min_length": 37.0, | |
| "eval_completions/min_terminated_length": 37.0, | |
| "eval_loss": 0.0, | |
| "eval_num_tokens": 17189416.0, | |
| "eval_reward": 10.357580423355103, | |
| "eval_reward_std": 0.5738533288240433, | |
| "eval_rewards/accuracy_reward": 0.76953125, | |
| "eval_rewards/brier_reward": 0.8266183733940125, | |
| "eval_rewards/confidence_one_or_zero": 0.0, | |
| "eval_rewards/format_reward": 1.0, | |
| "eval_rewards/log_2_reward": 9.472814798355103, | |
| "eval_rewards/mean_confidence_reward": 0.7106640785932541, | |
| "eval_runtime": 6.4379, | |
| "eval_samples_per_second": 77.665, | |
| "eval_steps_per_second": 0.621, | |
| "step": 50 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.0, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 37.971484375, | |
| "completions/mean_terminated_length": 37.971484375, | |
| "completions/min_length": 34.6, | |
| "completions/min_terminated_length": 34.6, | |
| "epoch": 0.3352380952380952, | |
| "grad_norm": 0.0016840703319758177, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 18888964.0, | |
| "reward": 10.318778038024902, | |
| "reward_std": 0.02350865714251995, | |
| "rewards/accuracy_reward": 0.74375, | |
| "rewards/brier_reward": 0.8148023843765259, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.446902847290039, | |
| "rewards/mean_confidence_reward": 0.715139651298523, | |
| "step": 55 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.0, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 37.99951171875, | |
| "completions/mean_terminated_length": 37.99951171875, | |
| "completions/min_length": 34.4, | |
| "completions/min_terminated_length": 34.4, | |
| "epoch": 0.3657142857142857, | |
| "grad_norm": 0.0030926617328077555, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 20588799.0, | |
| "reward": 10.39038314819336, | |
| "reward_std": 0.022050340101122855, | |
| "rewards/accuracy_reward": 0.790625, | |
| "rewards/brier_reward": 0.8362791776657105, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.495070266723634, | |
| "rewards/mean_confidence_reward": 0.7193691492080688, | |
| "step": 60 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.0, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 38.00595703125, | |
| "completions/mean_terminated_length": 38.00595703125, | |
| "completions/min_length": 34.4, | |
| "completions/min_terminated_length": 34.4, | |
| "epoch": 0.3961904761904762, | |
| "grad_norm": 0.002488529309630394, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 22288700.0, | |
| "reward": 10.282398986816407, | |
| "reward_std": 0.026306112483143806, | |
| "rewards/accuracy_reward": 0.71875, | |
| "rewards/brier_reward": 0.8037564754486084, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.423023986816407, | |
| "rewards/mean_confidence_reward": 0.715283215045929, | |
| "step": 65 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.0, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 38.00537109375, | |
| "completions/mean_terminated_length": 38.00537109375, | |
| "completions/min_length": 33.6, | |
| "completions/min_terminated_length": 33.6, | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 0.0028004285413771868, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 23988595.0, | |
| "reward": 10.350898361206054, | |
| "reward_std": 0.031046164222061634, | |
| "rewards/accuracy_reward": 0.76875, | |
| "rewards/brier_reward": 0.8238465666770936, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.466522979736329, | |
| "rewards/mean_confidence_reward": 0.7099578261375428, | |
| "step": 70 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.009765625, | |
| "completions/mean_terminated_length": 38.009765625, | |
| "completions/min_length": 34.6, | |
| "completions/min_terminated_length": 34.6, | |
| "epoch": 0.45714285714285713, | |
| "grad_norm": 0.005292444955557585, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 25688535.0, | |
| "reward": 10.329157447814941, | |
| "reward_std": 0.036872823536396024, | |
| "rewards/accuracy_reward": 0.74375, | |
| "rewards/brier_reward": 0.81919184923172, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.457282257080077, | |
| "rewards/mean_confidence_reward": 0.7013833165168762, | |
| "step": 75 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.0, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 37.9822265625, | |
| "completions/mean_terminated_length": 37.9822265625, | |
| "completions/min_length": 34.6, | |
| "completions/min_terminated_length": 34.6, | |
| "epoch": 0.4876190476190476, | |
| "grad_norm": 0.004484053235501051, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 27388193.0, | |
| "reward": 10.357734870910644, | |
| "reward_std": 0.03486044779419899, | |
| "rewards/accuracy_reward": 0.765625, | |
| "rewards/brier_reward": 0.8267481923103333, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.474921989440919, | |
| "rewards/mean_confidence_reward": 0.7021149516105651, | |
| "step": 80 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.0, | |
| "completions/max_terminated_length": 42.0, | |
| "completions/mean_length": 37.94638671875, | |
| "completions/mean_terminated_length": 37.94638671875, | |
| "completions/min_length": 33.8, | |
| "completions/min_terminated_length": 33.8, | |
| "epoch": 0.518095238095238, | |
| "grad_norm": 0.0053267451003193855, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 29087484.0, | |
| "reward": 10.338480186462402, | |
| "reward_std": 0.027601083368062974, | |
| "rewards/accuracy_reward": 0.740625, | |
| "rewards/brier_reward": 0.8229233026504517, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.468167877197265, | |
| "rewards/mean_confidence_reward": 0.7042033433914184, | |
| "step": 85 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.2, | |
| "completions/max_terminated_length": 41.2, | |
| "completions/mean_length": 37.9126953125, | |
| "completions/mean_terminated_length": 37.9126953125, | |
| "completions/min_length": 33.8, | |
| "completions/min_terminated_length": 33.8, | |
| "epoch": 0.5485714285714286, | |
| "grad_norm": 0.0024137713480740786, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 30786430.0, | |
| "reward": 10.364272308349609, | |
| "reward_std": 0.019712066650390624, | |
| "rewards/accuracy_reward": 0.771875, | |
| "rewards/brier_reward": 0.828098464012146, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.47833480834961, | |
| "rewards/mean_confidence_reward": 0.7173008680343628, | |
| "step": 90 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.2, | |
| "completions/max_terminated_length": 41.2, | |
| "completions/mean_length": 37.84951171875, | |
| "completions/mean_terminated_length": 37.84951171875, | |
| "completions/min_length": 33.0, | |
| "completions/min_terminated_length": 33.0, | |
| "epoch": 0.579047619047619, | |
| "grad_norm": 0.0020697573199868202, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 32484729.0, | |
| "reward": 10.36720905303955, | |
| "reward_std": 0.022712503001093863, | |
| "rewards/accuracy_reward": 0.771875, | |
| "rewards/brier_reward": 0.8293002605438232, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.481271362304687, | |
| "rewards/mean_confidence_reward": 0.7178742527961731, | |
| "step": 95 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.2, | |
| "completions/max_terminated_length": 41.2, | |
| "completions/mean_length": 37.87900390625, | |
| "completions/mean_terminated_length": 37.87900390625, | |
| "completions/min_length": 34.0, | |
| "completions/min_terminated_length": 34.0, | |
| "epoch": 0.6095238095238096, | |
| "grad_norm": 0.0031165352556854486, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 34183330.0, | |
| "reward": 10.337306022644043, | |
| "reward_std": 0.023966287076473237, | |
| "rewards/accuracy_reward": 0.734375, | |
| "rewards/brier_reward": 0.8224358916282654, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.470118141174316, | |
| "rewards/mean_confidence_reward": 0.7064209103584289, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.6095238095238096, | |
| "eval_completions/clipped_ratio": 0.0, | |
| "eval_completions/max_length": 38.0, | |
| "eval_completions/max_terminated_length": 38.0, | |
| "eval_completions/mean_length": 37.89486789703369, | |
| "eval_completions/mean_terminated_length": 37.89486789703369, | |
| "eval_completions/min_length": 37.0, | |
| "eval_completions/min_terminated_length": 37.0, | |
| "eval_loss": 0.0, | |
| "eval_num_tokens": 34183330.0, | |
| "eval_reward": 10.359020471572876, | |
| "eval_reward_std": 0.5875897854566574, | |
| "eval_rewards/accuracy_reward": 0.76953125, | |
| "eval_rewards/brier_reward": 0.8270021378993988, | |
| "eval_rewards/confidence_one_or_zero": 0.0, | |
| "eval_rewards/format_reward": 1.0, | |
| "eval_rewards/log_2_reward": 9.474255084991455, | |
| "eval_rewards/mean_confidence_reward": 0.7051367163658142, | |
| "eval_runtime": 5.8203, | |
| "eval_samples_per_second": 85.906, | |
| "eval_steps_per_second": 0.687, | |
| "step": 100 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 39.6, | |
| "completions/max_terminated_length": 39.6, | |
| "completions/mean_length": 37.93349609375, | |
| "completions/mean_terminated_length": 37.93349609375, | |
| "completions/min_length": 33.8, | |
| "completions/min_terminated_length": 33.8, | |
| "epoch": 0.64, | |
| "grad_norm": 0.00220179813914001, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 35882489.0, | |
| "reward": 10.314545440673829, | |
| "reward_std": 0.029955898225307465, | |
| "rewards/accuracy_reward": 0.73125, | |
| "rewards/brier_reward": 0.8140996694564819, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.448920059204102, | |
| "rewards/mean_confidence_reward": 0.7027922987937927, | |
| "step": 105 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 40.2, | |
| "completions/max_terminated_length": 40.2, | |
| "completions/mean_length": 37.9857421875, | |
| "completions/mean_terminated_length": 37.9857421875, | |
| "completions/min_length": 37.0, | |
| "completions/min_terminated_length": 37.0, | |
| "epoch": 0.6704761904761904, | |
| "grad_norm": 0.002244519302621484, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 37582183.0, | |
| "reward": 10.317130851745606, | |
| "reward_std": 0.02459430545568466, | |
| "rewards/accuracy_reward": 0.725, | |
| "rewards/brier_reward": 0.8168185353279114, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.454631042480468, | |
| "rewards/mean_confidence_reward": 0.6982184529304505, | |
| "step": 110 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.00478515625, | |
| "completions/mean_terminated_length": 38.00478515625, | |
| "completions/min_length": 35.2, | |
| "completions/min_terminated_length": 35.2, | |
| "epoch": 0.700952380952381, | |
| "grad_norm": 0.005517852492630482, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 39282072.0, | |
| "reward": 10.32682113647461, | |
| "reward_std": 0.027102234587073327, | |
| "rewards/accuracy_reward": 0.74375, | |
| "rewards/brier_reward": 0.81859370470047, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.454946327209473, | |
| "rewards/mean_confidence_reward": 0.7090798854827881, | |
| "step": 115 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.4, | |
| "completions/max_terminated_length": 41.4, | |
| "completions/mean_length": 38.005078125, | |
| "completions/mean_terminated_length": 38.005078125, | |
| "completions/min_length": 35.2, | |
| "completions/min_terminated_length": 35.2, | |
| "epoch": 0.7314285714285714, | |
| "grad_norm": 0.001279509044252336, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 40981964.0, | |
| "reward": 10.33778247833252, | |
| "reward_std": 0.016821843385696412, | |
| "rewards/accuracy_reward": 0.740625, | |
| "rewards/brier_reward": 0.8214999318122864, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.467469787597656, | |
| "rewards/mean_confidence_reward": 0.7188453316688538, | |
| "step": 120 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.02109375, | |
| "completions/mean_terminated_length": 38.02109375, | |
| "completions/min_length": 36.4, | |
| "completions/min_terminated_length": 36.4, | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 0.004740776959806681, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 42682020.0, | |
| "reward": 10.347677993774415, | |
| "reward_std": 0.014656592346727847, | |
| "rewards/accuracy_reward": 0.746875, | |
| "rewards/brier_reward": 0.8234950304031372, | |
| "rewards/confidence_one_or_zero": 0.0, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.474240493774413, | |
| "rewards/mean_confidence_reward": 0.725271201133728, | |
| "step": 125 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.02802734375, | |
| "completions/mean_terminated_length": 38.02802734375, | |
| "completions/min_length": 35.8, | |
| "completions/min_terminated_length": 35.8, | |
| "epoch": 0.7923809523809524, | |
| "grad_norm": 0.0035361009649932384, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 44382147.0, | |
| "reward": 10.308923149108887, | |
| "reward_std": 0.015245725214481354, | |
| "rewards/accuracy_reward": 0.715625, | |
| "rewards/brier_reward": 0.8122243165969849, | |
| "rewards/confidence_one_or_zero": 0.00224609375, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.451111030578613, | |
| "rewards/mean_confidence_reward": 0.7021763563156128, | |
| "step": 130 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.0369140625, | |
| "completions/mean_terminated_length": 38.0369140625, | |
| "completions/min_length": 34.4, | |
| "completions/min_terminated_length": 34.4, | |
| "epoch": 0.8228571428571428, | |
| "grad_norm": 0.0022523615043610334, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 46082365.0, | |
| "reward": 10.33953399658203, | |
| "reward_std": 0.009829406999051571, | |
| "rewards/accuracy_reward": 0.746875, | |
| "rewards/brier_reward": 0.8228122234344483, | |
| "rewards/confidence_one_or_zero": 0.01396484375, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.466096305847168, | |
| "rewards/mean_confidence_reward": 0.7044583916664123, | |
| "step": 135 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.0568359375, | |
| "completions/mean_terminated_length": 38.0568359375, | |
| "completions/min_length": 36.2, | |
| "completions/min_terminated_length": 36.2, | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 0.0014996592653915286, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 47782787.0, | |
| "reward": 10.351680374145507, | |
| "reward_std": 0.010882622562348843, | |
| "rewards/accuracy_reward": 0.765625, | |
| "rewards/brier_reward": 0.8265089988708496, | |
| "rewards/confidence_one_or_zero": 0.0072265625, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.468867874145507, | |
| "rewards/mean_confidence_reward": 0.7003974556922913, | |
| "step": 140 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 40.2, | |
| "completions/max_terminated_length": 40.2, | |
| "completions/mean_length": 38.04287109375, | |
| "completions/mean_terminated_length": 38.04287109375, | |
| "completions/min_length": 36.6, | |
| "completions/min_terminated_length": 36.6, | |
| "epoch": 0.8838095238095238, | |
| "grad_norm": 0.0019739060662686825, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 49483066.0, | |
| "reward": 10.277253150939941, | |
| "reward_std": 0.011264503002166748, | |
| "rewards/accuracy_reward": 0.709375, | |
| "rewards/brier_reward": 0.803178334236145, | |
| "rewards/confidence_one_or_zero": 0.0056640625, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.422565841674805, | |
| "rewards/mean_confidence_reward": 0.6882402420043945, | |
| "step": 145 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.6, | |
| "completions/max_terminated_length": 41.6, | |
| "completions/mean_length": 38.0552734375, | |
| "completions/mean_terminated_length": 38.0552734375, | |
| "completions/min_length": 35.4, | |
| "completions/min_terminated_length": 35.4, | |
| "epoch": 0.9142857142857143, | |
| "grad_norm": 0.002162984339520335, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 51183472.0, | |
| "reward": 10.361213111877442, | |
| "reward_std": 0.009932122565805912, | |
| "rewards/accuracy_reward": 0.765625, | |
| "rewards/brier_reward": 0.8259004116058349, | |
| "rewards/confidence_one_or_zero": 0.0162109375, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.478400802612304, | |
| "rewards/mean_confidence_reward": 0.7085066318511963, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.9142857142857143, | |
| "eval_completions/clipped_ratio": 0.0, | |
| "eval_completions/max_length": 39.0, | |
| "eval_completions/max_terminated_length": 39.0, | |
| "eval_completions/mean_length": 38.07933712005615, | |
| "eval_completions/mean_terminated_length": 38.07933712005615, | |
| "eval_completions/min_length": 38.0, | |
| "eval_completions/min_terminated_length": 38.0, | |
| "eval_loss": 0.0, | |
| "eval_num_tokens": 51183472.0, | |
| "eval_reward": 10.357120752334595, | |
| "eval_reward_std": 0.6193633526563644, | |
| "eval_rewards/accuracy_reward": 0.76953125, | |
| "eval_rewards/brier_reward": 0.8273185789585114, | |
| "eval_rewards/confidence_one_or_zero": 0.009765625, | |
| "eval_rewards/format_reward": 1.0, | |
| "eval_rewards/log_2_reward": 9.472355365753174, | |
| "eval_rewards/mean_confidence_reward": 0.7049629092216492, | |
| "eval_runtime": 5.8271, | |
| "eval_samples_per_second": 85.807, | |
| "eval_steps_per_second": 0.686, | |
| "step": 150 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 40.8, | |
| "completions/max_terminated_length": 40.8, | |
| "completions/mean_length": 38.05673828125, | |
| "completions/mean_terminated_length": 38.05673828125, | |
| "completions/min_length": 35.4, | |
| "completions/min_terminated_length": 35.4, | |
| "epoch": 0.9447619047619048, | |
| "grad_norm": 0.002443034201860428, | |
| "learning_rate": 1e-06, | |
| "loss": -0.0, | |
| "num_tokens": 52883893.0, | |
| "reward": 10.319277572631837, | |
| "reward_std": 0.012516670487821102, | |
| "rewards/accuracy_reward": 0.740625, | |
| "rewards/brier_reward": 0.8164864778518677, | |
| "rewards/confidence_one_or_zero": 0.0041015625, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.448964500427246, | |
| "rewards/mean_confidence_reward": 0.6974437594413757, | |
| "step": 155 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 42.2, | |
| "completions/max_terminated_length": 42.2, | |
| "completions/mean_length": 38.0748046875, | |
| "completions/mean_terminated_length": 38.0748046875, | |
| "completions/min_length": 35.8, | |
| "completions/min_terminated_length": 35.8, | |
| "epoch": 0.9752380952380952, | |
| "grad_norm": 0.004797317553311586, | |
| "learning_rate": 1e-06, | |
| "loss": 0.0, | |
| "num_tokens": 54584499.0, | |
| "reward": 10.403311920166015, | |
| "reward_std": 0.02263805828988552, | |
| "rewards/accuracy_reward": 0.775, | |
| "rewards/brier_reward": 0.8423254013061523, | |
| "rewards/confidence_one_or_zero": 0.0095703125, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.515811729431153, | |
| "rewards/mean_confidence_reward": 0.7259237408638001, | |
| "step": 160 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0, | |
| "completions/max_length": 41.25, | |
| "completions/max_terminated_length": 41.25, | |
| "completions/mean_length": 38.05098247528076, | |
| "completions/mean_terminated_length": 38.05098247528076, | |
| "completions/min_length": 36.5, | |
| "completions/min_terminated_length": 36.5, | |
| "epoch": 0.9996190476190476, | |
| "num_tokens": 55944868.0, | |
| "reward": 10.392438888549805, | |
| "reward_std": 0.01846723770722747, | |
| "rewards/accuracy_reward": 0.7578125, | |
| "rewards/brier_reward": 0.8418954759836197, | |
| "rewards/confidence_one_or_zero": 0.006591796875, | |
| "rewards/format_reward": 1.0, | |
| "rewards/log_2_reward": 9.513532400131226, | |
| "rewards/mean_confidence_reward": 0.7270061075687408, | |
| "step": 164, | |
| "total_flos": 0.0, | |
| "train_loss": 0.009617539934250022, | |
| "train_runtime": 10848.9507, | |
| "train_samples_per_second": 0.968, | |
| "train_steps_per_second": 0.015 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 164, | |
| "num_input_tokens_seen": 55944868, | |
| "num_train_epochs": 1, | |
| "save_steps": 60, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |