toy-log-RLCR-v1 / trainer_state.json
mehuldamani's picture
Model save
923b2f2 verified
{
"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": 0.24140625,
"completions/max_length": 64.6,
"completions/max_terminated_length": 64.4,
"completions/mean_length": 63.24580078125,
"completions/mean_terminated_length": 63.16332855224609,
"completions/min_length": 16.2,
"completions/min_terminated_length": 16.2,
"epoch": 0.030476190476190476,
"grad_norm": 0.4937044382095337,
"learning_rate": 5.555555555555555e-07,
"loss": 0.0283,
"num_tokens": 1958357.0,
"reward": 0.147213414311409,
"reward_std": 0.6661659717559815,
"rewards/accuracy_reward": 0.01796875,
"rewards/brier_reward": 0.02096688412129879,
"rewards/confidence_one_or_zero": 0.0072265625,
"rewards/format_reward": 0.0279296875,
"rewards/log_2_reward": 0.24852838814258577,
"rewards/mean_confidence_reward": 0.03470296487212181,
"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": 0.366015625,
"completions/max_length": 64.6,
"completions/max_terminated_length": 64.4,
"completions/mean_length": 61.88505859375,
"completions/mean_terminated_length": 60.428192138671875,
"completions/min_length": 11.2,
"completions/min_terminated_length": 11.2,
"epoch": 0.06095238095238095,
"grad_norm": 0.24706685543060303,
"learning_rate": 1e-06,
"loss": 0.0898,
"num_tokens": 3902780.0,
"reward": 0.45152418315410614,
"reward_std": 1.2864935278892518,
"rewards/accuracy_reward": 0.0533203125,
"rewards/brier_reward": 0.06290790475904942,
"rewards/confidence_one_or_zero": 0.02109375,
"rewards/format_reward": 0.0875,
"rewards/log_2_reward": 0.7622280478477478,
"rewards/mean_confidence_reward": 0.08969703018665313,
"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": 0.0689453125,
"completions/max_length": 64.2,
"completions/max_terminated_length": 64.2,
"completions/mean_length": 47.2318359375,
"completions/mean_terminated_length": 46.269396209716795,
"completions/min_length": 17.6,
"completions/min_terminated_length": 17.6,
"epoch": 0.09142857142857143,
"grad_norm": 0.0876019150018692,
"learning_rate": 1e-06,
"loss": 0.2805,
"num_tokens": 5697154.0,
"reward": 3.8407562732696534,
"reward_std": 2.1547468423843386,
"rewards/accuracy_reward": 0.47509765625,
"rewards/brier_reward": 0.5359858989715576,
"rewards/confidence_one_or_zero": 0.1521484375,
"rewards/format_reward": 0.7376953125,
"rewards/log_2_reward": 6.46871919631958,
"rewards/mean_confidence_reward": 0.6385979354381561,
"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": 0.00185546875,
"completions/max_length": 63.4,
"completions/max_terminated_length": 61.6,
"completions/mean_length": 37.20126953125,
"completions/mean_terminated_length": 37.15373306274414,
"completions/min_length": 19.6,
"completions/min_terminated_length": 19.6,
"epoch": 0.1219047619047619,
"grad_norm": 0.01609966531395912,
"learning_rate": 1e-06,
"loss": 0.0059,
"num_tokens": 7388815.0,
"reward": 5.555976676940918,
"reward_std": 0.33366540521383287,
"rewards/accuracy_reward": 0.71943359375,
"rewards/brier_reward": 0.8196830987930298,
"rewards/confidence_one_or_zero": 0.07275390625,
"rewards/format_reward": 0.9970703125,
"rewards/log_2_reward": 9.395448875427245,
"rewards/mean_confidence_reward": 0.7144028902053833,
"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": 0.0,
"completions/max_length": 54.0,
"completions/max_terminated_length": 54.0,
"completions/mean_length": 35.7521484375,
"completions/mean_terminated_length": 35.7521484375,
"completions/min_length": 28.8,
"completions/min_terminated_length": 28.8,
"epoch": 0.1523809523809524,
"grad_norm": 0.007558116689324379,
"learning_rate": 1e-06,
"loss": -0.0006,
"num_tokens": 9065637.0,
"reward": 5.614911460876465,
"reward_std": 0.0848025493323803,
"rewards/accuracy_reward": 0.7482421875,
"rewards/brier_reward": 0.8309232592582703,
"rewards/confidence_one_or_zero": 0.02490234375,
"rewards/format_reward": 0.99921875,
"rewards/log_2_reward": 9.482362365722656,
"rewards/mean_confidence_reward": 0.7317079305648804,
"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": 0.0,
"completions/max_length": 51.4,
"completions/max_terminated_length": 51.4,
"completions/mean_length": 34.6220703125,
"completions/mean_terminated_length": 34.6220703125,
"completions/min_length": 32.4,
"completions/min_terminated_length": 32.4,
"epoch": 0.18285714285714286,
"grad_norm": 0.0032027510460466146,
"learning_rate": 1e-06,
"loss": -0.0001,
"num_tokens": 10730887.0,
"reward": 5.600866794586182,
"reward_std": 0.02974188979715109,
"rewards/accuracy_reward": 0.71484375,
"rewards/brier_reward": 0.831008231639862,
"rewards/confidence_one_or_zero": 0.01875,
"rewards/format_reward": 0.99990234375,
"rewards/log_2_reward": 9.48698787689209,
"rewards/mean_confidence_reward": 0.7338953256607056,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 36.87431640625,
"completions/mean_terminated_length": 36.87431640625,
"completions/min_length": 32.2,
"completions/min_terminated_length": 32.2,
"epoch": 0.21333333333333335,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 12419200.0,
"reward": 5.571425914764404,
"reward_std": 0.005977154635274928,
"rewards/accuracy_reward": 0.71533203125,
"rewards/brier_reward": 0.807461929321289,
"rewards/confidence_one_or_zero": 0.021875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.427520751953125,
"rewards/mean_confidence_reward": 0.756035315990448,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 37.90732421875,
"completions/mean_terminated_length": 37.90732421875,
"completions/min_length": 33.4,
"completions/min_terminated_length": 33.4,
"epoch": 0.2438095238095238,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 14118091.0,
"reward": 5.616148853302002,
"reward_std": 3.089594500238491e-08,
"rewards/accuracy_reward": 0.75,
"rewards/brier_reward": 0.8339757204055787,
"rewards/confidence_one_or_zero": 0.021875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.482297897338867,
"rewards/mean_confidence_reward": 0.7595968842506409,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 37.988671875,
"completions/mean_terminated_length": 37.988671875,
"completions/min_length": 34.2,
"completions/min_terminated_length": 34.2,
"epoch": 0.2742857142857143,
"grad_norm": 1.1408615563368585e-07,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 15817815.0,
"reward": 5.657996559143067,
"reward_std": 8.469001613775617e-07,
"rewards/accuracy_reward": 0.7875,
"rewards/brier_reward": 0.8462499141693115,
"rewards/confidence_one_or_zero": 0.021875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.528493499755859,
"rewards/mean_confidence_reward": 0.7478750228881836,
"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": 0.0,
"completions/max_length": 51.4,
"completions/max_terminated_length": 51.4,
"completions/mean_length": 37.9708984375,
"completions/mean_terminated_length": 37.9708984375,
"completions/min_length": 34.6,
"completions/min_terminated_length": 34.6,
"epoch": 0.3047619047619048,
"grad_norm": 5.573377848122618e-07,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 17517357.0,
"reward": 5.583265399932861,
"reward_std": 0.000277167129360123,
"rewards/accuracy_reward": 0.72177734375,
"rewards/brier_reward": 0.8111650466918945,
"rewards/confidence_one_or_zero": 0.009375,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.444753646850586,
"rewards/mean_confidence_reward": 0.756777310371399,
"step": 50
},
{
"epoch": 0.3047619047619048,
"eval_completions/clipped_ratio": 0.0,
"eval_completions/max_length": 47.75,
"eval_completions/max_terminated_length": 47.75,
"eval_completions/mean_length": 38.02060890197754,
"eval_completions/mean_terminated_length": 38.02060890197754,
"eval_completions/min_length": 36.0,
"eval_completions/min_terminated_length": 36.0,
"eval_loss": 0.0,
"eval_num_tokens": 17517357.0,
"eval_reward": 5.624788165092468,
"eval_reward_std": 0.46847544610500336,
"eval_rewards/accuracy_reward": 0.76953125,
"eval_rewards/brier_reward": 0.8338424563407898,
"eval_rewards/confidence_one_or_zero": 0.021484375,
"eval_rewards/format_reward": 1.0,
"eval_rewards/log_2_reward": 9.480044841766357,
"eval_rewards/mean_confidence_reward": 0.7464218735694885,
"eval_runtime": 6.6713,
"eval_samples_per_second": 74.948,
"eval_steps_per_second": 0.6,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.0583984375,
"completions/mean_terminated_length": 38.0583984375,
"completions/min_length": 35.0,
"completions/min_terminated_length": 35.0,
"epoch": 0.3352380952380952,
"grad_norm": 2.0944073142459274e-08,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 19217795.0,
"reward": 5.633004188537598,
"reward_std": 4.244793103680422e-08,
"rewards/accuracy_reward": 0.74375,
"rewards/brier_reward": 0.8407172799110413,
"rewards/confidence_one_or_zero": 0.025,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.522258758544922,
"rewards/mean_confidence_reward": 0.7552843928337097,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 38.00673828125,
"completions/mean_terminated_length": 38.00673828125,
"completions/min_length": 35.0,
"completions/min_terminated_length": 35.0,
"epoch": 0.3657142857142857,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 20917704.0,
"reward": 5.66101713180542,
"reward_std": 0.0026967732763623076,
"rewards/accuracy_reward": 0.7875,
"rewards/brier_reward": 0.8482979655265808,
"rewards/confidence_one_or_zero": 0.0125,
"rewards/format_reward": 0.99990234375,
"rewards/log_2_reward": 9.534632110595703,
"rewards/mean_confidence_reward": 0.7581686615943909,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 38.03154296875,
"completions/mean_terminated_length": 38.03154296875,
"completions/min_length": 35.6,
"completions/min_terminated_length": 35.6,
"epoch": 0.3961904761904762,
"grad_norm": 9.332956096841372e-08,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 22617867.0,
"reward": 5.594816780090332,
"reward_std": 0.00027667832823539127,
"rewards/accuracy_reward": 0.72197265625,
"rewards/brier_reward": 0.8180258274078369,
"rewards/confidence_one_or_zero": 0.025,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.467660713195801,
"rewards/mean_confidence_reward": 0.7601582884788514,
"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": 9.765625e-05,
"completions/max_length": 53.8,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 38.0232421875,
"completions/mean_terminated_length": 38.020800018310545,
"completions/min_length": 35.6,
"completions/min_terminated_length": 35.6,
"epoch": 0.4266666666666667,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0002,
"num_tokens": 24317945.0,
"reward": 5.6432279586792,
"reward_std": 0.0026955028995871546,
"rewards/accuracy_reward": 0.76875,
"rewards/brier_reward": 0.8412735939025879,
"rewards/confidence_one_or_zero": 0.009375,
"rewards/format_reward": 0.99990234375,
"rewards/log_2_reward": 9.517803573608399,
"rewards/mean_confidence_reward": 0.7524774312973023,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 38.0142578125,
"completions/mean_terminated_length": 38.0142578125,
"completions/min_length": 36.6,
"completions/min_terminated_length": 36.6,
"epoch": 0.45714285714285713,
"grad_norm": 2.269921139941289e-08,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 26017931.0,
"reward": 5.630081748962402,
"reward_std": 4.863502539365072e-07,
"rewards/accuracy_reward": 0.74375,
"rewards/brier_reward": 0.8380238056182862,
"rewards/confidence_one_or_zero": 0.00625,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.516414070129395,
"rewards/mean_confidence_reward": 0.7339439868927002,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 38.0216796875,
"completions/mean_terminated_length": 38.0216796875,
"completions/min_length": 36.2,
"completions/min_terminated_length": 36.2,
"epoch": 0.4876190476190476,
"grad_norm": 1.2988358832899394e-07,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 27717993.0,
"reward": 5.639082336425782,
"reward_std": 1.4793427283166238e-07,
"rewards/accuracy_reward": 0.759375,
"rewards/brier_reward": 0.8395630717277527,
"rewards/confidence_one_or_zero": 0.01884765625,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.51879005432129,
"rewards/mean_confidence_reward": 0.760356330871582,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.04814453125,
"completions/mean_terminated_length": 38.04814453125,
"completions/min_length": 35.8,
"completions/min_terminated_length": 35.8,
"epoch": 0.518095238095238,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 29418326.0,
"reward": 5.621638393402099,
"reward_std": 5.215314899942314e-07,
"rewards/accuracy_reward": 0.740625,
"rewards/brier_reward": 0.8350816965103149,
"rewards/confidence_one_or_zero": 0.015625,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.502651786804199,
"rewards/mean_confidence_reward": 0.7518718242645264,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.00283203125,
"completions/mean_terminated_length": 38.00283203125,
"completions/min_length": 36.4,
"completions/min_terminated_length": 36.4,
"epoch": 0.5485714285714286,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 31118195.0,
"reward": 5.64435043334961,
"reward_std": 9.689834064374736e-06,
"rewards/accuracy_reward": 0.771875,
"rewards/brier_reward": 0.8399171829223633,
"rewards/confidence_one_or_zero": 0.01875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.516826057434082,
"rewards/mean_confidence_reward": 0.7615166783332825,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 37.99736328125,
"completions/mean_terminated_length": 37.99736328125,
"completions/min_length": 36.6,
"completions/min_terminated_length": 36.6,
"epoch": 0.579047619047619,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 32818008.0,
"reward": 5.644574451446533,
"reward_std": 2.637261195559404e-06,
"rewards/accuracy_reward": 0.771875,
"rewards/brier_reward": 0.8405626773834228,
"rewards/confidence_one_or_zero": 0.0125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.517273902893066,
"rewards/mean_confidence_reward": 0.7512939691543579,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 37.9896484375,
"completions/mean_terminated_length": 37.9896484375,
"completions/min_length": 36.0,
"completions/min_terminated_length": 36.0,
"epoch": 0.6095238095238096,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 34517742.0,
"reward": 5.623265933990479,
"reward_std": 2.0761127643709188e-07,
"rewards/accuracy_reward": 0.7375,
"rewards/brier_reward": 0.8349212646484375,
"rewards/confidence_one_or_zero": 0.040625,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.509031677246094,
"rewards/mean_confidence_reward": 0.7413062334060669,
"step": 100
},
{
"epoch": 0.6095238095238096,
"eval_completions/clipped_ratio": 0.0,
"eval_completions/max_length": 41.5,
"eval_completions/max_terminated_length": 41.5,
"eval_completions/mean_length": 37.94639015197754,
"eval_completions/mean_terminated_length": 37.94639015197754,
"eval_completions/min_length": 36.0,
"eval_completions/min_terminated_length": 36.0,
"eval_loss": 0.0,
"eval_num_tokens": 34517742.0,
"eval_reward": 5.625764727592468,
"eval_reward_std": 0.46847712993621826,
"eval_rewards/accuracy_reward": 0.771484375,
"eval_rewards/brier_reward": 0.8338424563407898,
"eval_rewards/confidence_one_or_zero": 0.021484375,
"eval_rewards/format_reward": 1.0,
"eval_rewards/log_2_reward": 9.480044841766357,
"eval_rewards/mean_confidence_reward": 0.7464453130960464,
"eval_runtime": 6.321,
"eval_samples_per_second": 79.101,
"eval_steps_per_second": 0.633,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.013671875,
"completions/mean_terminated_length": 38.013671875,
"completions/min_length": 36.6,
"completions/min_terminated_length": 36.6,
"epoch": 0.64,
"grad_norm": 2.458902770285931e-07,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 36217722.0,
"reward": 5.606796550750732,
"reward_std": 4.680077182683817e-07,
"rewards/accuracy_reward": 0.73125,
"rewards/brier_reward": 0.8244146585464478,
"rewards/confidence_one_or_zero": 0.021875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.482342720031738,
"rewards/mean_confidence_reward": 0.7441876769065857,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 38.02109375,
"completions/mean_terminated_length": 38.02109375,
"completions/min_length": 35.4,
"completions/min_terminated_length": 35.4,
"epoch": 0.6704761904761904,
"grad_norm": 4.270423517027666e-07,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 37917778.0,
"reward": 5.600750255584717,
"reward_std": 1.2465623044022323e-06,
"rewards/accuracy_reward": 0.725,
"rewards/brier_reward": 0.8319555759429932,
"rewards/confidence_one_or_zero": 0.00625,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.476500701904296,
"rewards/mean_confidence_reward": 0.7388657212257386,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.03330078125,
"completions/mean_terminated_length": 38.03330078125,
"completions/min_length": 36.4,
"completions/min_terminated_length": 36.4,
"epoch": 0.700952380952381,
"grad_norm": 1.4557041794205361e-08,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 39617959.0,
"reward": 5.6239110946655275,
"reward_std": 1.9214322506400093e-07,
"rewards/accuracy_reward": 0.74375,
"rewards/brier_reward": 0.8344948053359985,
"rewards/confidence_one_or_zero": 0.028125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.504072189331055,
"rewards/mean_confidence_reward": 0.761590576171875,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 37.98671875,
"completions/mean_terminated_length": 37.98671875,
"completions/min_length": 35.8,
"completions/min_terminated_length": 35.8,
"epoch": 0.7314285714285714,
"grad_norm": 3.9047936439828845e-08,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 41317663.0,
"reward": 5.612829780578613,
"reward_std": 5.51469579335162e-08,
"rewards/accuracy_reward": 0.740625,
"rewards/brier_reward": 0.8286247611045837,
"rewards/confidence_one_or_zero": 0.0125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.48503475189209,
"rewards/mean_confidence_reward": 0.7615031599998474,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 38.0029296875,
"completions/mean_terminated_length": 38.0029296875,
"completions/min_length": 36.2,
"completions/min_terminated_length": 36.2,
"epoch": 0.7619047619047619,
"grad_norm": 9.443181170354364e-08,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 43017533.0,
"reward": 5.626171684265136,
"reward_std": 5.714372832699155e-07,
"rewards/accuracy_reward": 0.74375,
"rewards/brier_reward": 0.834185004234314,
"rewards/confidence_one_or_zero": 0.01875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.50859317779541,
"rewards/mean_confidence_reward": 0.768284547328949,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 38.01806640625,
"completions/mean_terminated_length": 38.01806640625,
"completions/min_length": 36.6,
"completions/min_terminated_length": 36.6,
"epoch": 0.7923809523809524,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 44717558.0,
"reward": 5.594818878173828,
"reward_std": 2.063908056015862e-07,
"rewards/accuracy_reward": 0.71875,
"rewards/brier_reward": 0.8205490350723267,
"rewards/confidence_one_or_zero": 0.028125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.47088794708252,
"rewards/mean_confidence_reward": 0.7437156438827515,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.01826171875,
"completions/mean_terminated_length": 38.01826171875,
"completions/min_length": 36.8,
"completions/min_terminated_length": 36.8,
"epoch": 0.8228571428571428,
"grad_norm": 0.002601601416245103,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 46417585.0,
"reward": 5.612351322174073,
"reward_std": 0.0027639744678197077,
"rewards/accuracy_reward": 0.746875,
"rewards/brier_reward": 0.8323373436927796,
"rewards/confidence_one_or_zero": 0.0375,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.47782859802246,
"rewards/mean_confidence_reward": 0.7525945901870728,
"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": 51.6,
"completions/max_terminated_length": 51.6,
"completions/mean_length": 38.02470703125,
"completions/mean_terminated_length": 38.02470703125,
"completions/min_length": 35.8,
"completions/min_terminated_length": 35.8,
"epoch": 0.8533333333333334,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 48117678.0,
"reward": 5.624975490570068,
"reward_std": 5.118799759173954e-07,
"rewards/accuracy_reward": 0.765625,
"rewards/brier_reward": 0.8380842685699463,
"rewards/confidence_one_or_zero": 0.028125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.484325790405274,
"rewards/mean_confidence_reward": 0.746293580532074,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 38.101171875,
"completions/mean_terminated_length": 38.101171875,
"completions/min_length": 36.8,
"completions/min_terminated_length": 36.8,
"epoch": 0.8838095238095238,
"grad_norm": 1.7589671941209417e-08,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 49818554.0,
"reward": 5.583425998687744,
"reward_std": 0.0031354847874361267,
"rewards/accuracy_reward": 0.70927734375,
"rewards/brier_reward": 0.81463543176651,
"rewards/confidence_one_or_zero": 0.021875,
"rewards/format_reward": 0.99990234375,
"rewards/log_2_reward": 9.457672691345214,
"rewards/mean_confidence_reward": 0.7387593507766723,
"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": 52.0,
"completions/max_terminated_length": 52.0,
"completions/mean_length": 38.06513671875,
"completions/mean_terminated_length": 38.06513671875,
"completions/min_length": 35.8,
"completions/min_terminated_length": 35.8,
"epoch": 0.9142857142857143,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 51519061.0,
"reward": 5.636477375030518,
"reward_std": 0.00027642484747048,
"rewards/accuracy_reward": 0.76552734375,
"rewards/brier_reward": 0.8363968968391419,
"rewards/confidence_one_or_zero": 0.034375,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.507427406311034,
"rewards/mean_confidence_reward": 0.7564831972122192,
"step": 150
},
{
"epoch": 0.9142857142857143,
"eval_completions/clipped_ratio": 0.0,
"eval_completions/max_length": 41.25,
"eval_completions/max_terminated_length": 41.25,
"eval_completions/mean_length": 37.91904640197754,
"eval_completions/mean_terminated_length": 37.91904640197754,
"eval_completions/min_length": 36.0,
"eval_completions/min_terminated_length": 36.0,
"eval_loss": 0.0,
"eval_num_tokens": 51519061.0,
"eval_reward": 5.625764727592468,
"eval_reward_std": 0.46847712993621826,
"eval_rewards/accuracy_reward": 0.771484375,
"eval_rewards/brier_reward": 0.8338424563407898,
"eval_rewards/confidence_one_or_zero": 0.021484375,
"eval_rewards/format_reward": 1.0,
"eval_rewards/log_2_reward": 9.480044841766357,
"eval_rewards/mean_confidence_reward": 0.7464453130960464,
"eval_runtime": 6.2742,
"eval_samples_per_second": 79.692,
"eval_steps_per_second": 0.638,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.05771484375,
"completions/mean_terminated_length": 38.05771484375,
"completions/min_length": 35.8,
"completions/min_terminated_length": 35.8,
"epoch": 0.9447619047619048,
"grad_norm": 1.0517731396930685e-07,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 53219492.0,
"reward": 5.621528244018554,
"reward_std": 8.822739161473692e-07,
"rewards/accuracy_reward": 0.740625,
"rewards/brier_reward": 0.8325549244880677,
"rewards/confidence_one_or_zero": 0.021875,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.502431297302246,
"rewards/mean_confidence_reward": 0.7421627283096314,
"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": 51.8,
"completions/max_terminated_length": 51.8,
"completions/mean_length": 38.035546875,
"completions/mean_terminated_length": 38.035546875,
"completions/min_length": 36.4,
"completions/min_terminated_length": 36.4,
"epoch": 0.9752380952380952,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": -0.0,
"num_tokens": 54919696.0,
"reward": 5.657506370544434,
"reward_std": 8.810078497845097e-08,
"rewards/accuracy_reward": 0.775,
"rewards/brier_reward": 0.8494255661964416,
"rewards/confidence_one_or_zero": 0.0125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.540012550354003,
"rewards/mean_confidence_reward": 0.7592812776565552,
"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": 51.75,
"completions/max_terminated_length": 51.75,
"completions/mean_length": 38.09580993652344,
"completions/mean_terminated_length": 38.09580993652344,
"completions/min_length": 36.75,
"completions/min_terminated_length": 36.75,
"epoch": 0.9996190476190476,
"num_tokens": 56280078.0,
"reward": 5.648329019546509,
"reward_std": 7.243279220858767e-08,
"rewards/accuracy_reward": 0.765625,
"rewards/brier_reward": 0.8462725430727005,
"rewards/confidence_one_or_zero": 0.03125,
"rewards/format_reward": 1.0,
"rewards/log_2_reward": 9.531033277511597,
"rewards/mean_confidence_reward": 0.7498086243867874,
"step": 164,
"total_flos": 0.0,
"train_loss": 0.012317530474541073,
"train_runtime": 10837.5902,
"train_samples_per_second": 0.969,
"train_steps_per_second": 0.015
}
],
"logging_steps": 5,
"max_steps": 164,
"num_input_tokens_seen": 56280078,
"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
}