2050 lines
80 KiB
JSON
2050 lines
80 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.6980392156862745,
|
|
"eval_steps": 500,
|
|
"global_step": 650,
|
|
"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": 0.0625,
|
|
"completions/max_length": 57.9,
|
|
"completions/max_terminated_length": 56.2,
|
|
"completions/mean_length": 28.353125,
|
|
"completions/mean_terminated_length": 26.12931594848633,
|
|
"completions/min_length": 4.6,
|
|
"completions/min_terminated_length": 4.6,
|
|
"entropy": 0.0189936759403281,
|
|
"epoch": 0.026143790849673203,
|
|
"frac_reward_zero_std": 0.91875,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.972307692307692e-06,
|
|
"loss": -0.0027778793126344683,
|
|
"num_tokens": 97161.0,
|
|
"reward": 0.8455517053604126,
|
|
"reward_std": 0.3738197267986834,
|
|
"rewards/reward_execution/mean": 0.8375,
|
|
"rewards/reward_execution/std": 0.30675073713064194,
|
|
"rewards/reward_format/mean": 0.00812500030733645,
|
|
"rewards/reward_format/std": 0.17115701287984847,
|
|
"rewards/reward_length/mean": -7.332145833061077e-05,
|
|
"rewards/reward_length/std": 0.0003982134105172008,
|
|
"step": 10,
|
|
"step_time": 30.141519616999993
|
|
},
|
|
{
|
|
"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": 0.109375,
|
|
"completions/max_length": 59.1,
|
|
"completions/max_terminated_length": 48.9,
|
|
"completions/mean_length": 28.325,
|
|
"completions/mean_terminated_length": 23.52349987030029,
|
|
"completions/min_length": 4.5,
|
|
"completions/min_terminated_length": 4.5,
|
|
"entropy": 0.014896306048285624,
|
|
"epoch": 0.05228758169934641,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.9415384615384614e-06,
|
|
"loss": -0.00027005113661289216,
|
|
"num_tokens": 190141.0,
|
|
"reward": 0.8436779797077179,
|
|
"reward_std": 0.4127595633268356,
|
|
"rewards/reward_execution/mean": 0.828125,
|
|
"rewards/reward_execution/std": 0.3320761829614639,
|
|
"rewards/reward_format/mean": 0.015625000465661287,
|
|
"rewards/reward_format/std": 0.16492117643356324,
|
|
"rewards/reward_length/mean": -7.204888370324625e-05,
|
|
"rewards/reward_length/std": 0.00022631439460383263,
|
|
"step": 20,
|
|
"step_time": 21.47389074279996
|
|
},
|
|
{
|
|
"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": 0.0875,
|
|
"completions/max_length": 58.7,
|
|
"completions/max_terminated_length": 54.8,
|
|
"completions/mean_length": 27.740625,
|
|
"completions/mean_terminated_length": 24.098472309112548,
|
|
"completions/min_length": 4.1,
|
|
"completions/min_terminated_length": 4.1,
|
|
"entropy": 0.022139274271603428,
|
|
"epoch": 0.0784313725490196,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.21031410992145538,
|
|
"learning_rate": 1.910769230769231e-06,
|
|
"loss": 0.001981881819665432,
|
|
"num_tokens": 283694.0,
|
|
"reward": 0.8389042556285858,
|
|
"reward_std": 0.3774765431880951,
|
|
"rewards/reward_execution/mean": 0.8234375,
|
|
"rewards/reward_execution/std": 0.3318361669778824,
|
|
"rewards/reward_format/mean": 0.015625000279396773,
|
|
"rewards/reward_format/std": 0.18242980688810348,
|
|
"rewards/reward_length/mean": -0.00015829465301067102,
|
|
"rewards/reward_length/std": 0.0008918488339986653,
|
|
"step": 30,
|
|
"step_time": 30.544650492100164
|
|
},
|
|
{
|
|
"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": 0.096875,
|
|
"completions/max_length": 60.4,
|
|
"completions/max_terminated_length": 53.8,
|
|
"completions/mean_length": 30.403125,
|
|
"completions/mean_terminated_length": 26.598913383483886,
|
|
"completions/min_length": 6.9,
|
|
"completions/min_terminated_length": 6.9,
|
|
"entropy": 0.014278956701309653,
|
|
"epoch": 0.10457516339869281,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.18257473409175873,
|
|
"learning_rate": 1.8799999999999998e-06,
|
|
"loss": 0.0035221416503190993,
|
|
"num_tokens": 378695.0,
|
|
"reward": 0.8087493270635605,
|
|
"reward_std": 0.4387156069278717,
|
|
"rewards/reward_execution/mean": 0.7765625,
|
|
"rewards/reward_execution/std": 0.3500940650701523,
|
|
"rewards/reward_format/mean": 0.03250000011175871,
|
|
"rewards/reward_format/std": 0.12955421656370164,
|
|
"rewards/reward_length/mean": -0.00031320829948526806,
|
|
"rewards/reward_length/std": 0.0014167540299240499,
|
|
"step": 40,
|
|
"step_time": 24.313605035499883
|
|
},
|
|
{
|
|
"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": 0.065625,
|
|
"completions/max_length": 60.8,
|
|
"completions/max_terminated_length": 56.0,
|
|
"completions/mean_length": 29.83125,
|
|
"completions/mean_terminated_length": 27.425152587890626,
|
|
"completions/min_length": 6.1,
|
|
"completions/min_terminated_length": 6.1,
|
|
"entropy": 0.02081742596656113,
|
|
"epoch": 0.13071895424836602,
|
|
"frac_reward_zero_std": 0.925,
|
|
"grad_norm": 0.18028853833675385,
|
|
"learning_rate": 1.8492307692307692e-06,
|
|
"loss": -0.00034371837973594664,
|
|
"num_tokens": 474349.0,
|
|
"reward": 0.8410520732402802,
|
|
"reward_std": 0.3922462262755289,
|
|
"rewards/reward_execution/mean": 0.83125,
|
|
"rewards/reward_execution/std": 0.315267439186573,
|
|
"rewards/reward_format/mean": 0.009999999776482582,
|
|
"rewards/reward_format/std": 0.1733264923095703,
|
|
"rewards/reward_length/mean": -0.0001979669468710199,
|
|
"rewards/reward_length/std": 0.0008465676353807793,
|
|
"step": 50,
|
|
"step_time": 29.170948042899887
|
|
},
|
|
{
|
|
"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": 0.153125,
|
|
"completions/max_length": 63.6,
|
|
"completions/max_terminated_length": 55.6,
|
|
"completions/mean_length": 34.4875,
|
|
"completions/mean_terminated_length": 29.20995502471924,
|
|
"completions/min_length": 6.4,
|
|
"completions/min_terminated_length": 6.4,
|
|
"entropy": 0.016935052816370443,
|
|
"epoch": 0.1568627450980392,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.13169364631175995,
|
|
"learning_rate": 1.8184615384615384e-06,
|
|
"loss": -0.0025871254503726957,
|
|
"num_tokens": 581457.0,
|
|
"reward": 0.8111869335174561,
|
|
"reward_std": 0.42564719915390015,
|
|
"rewards/reward_execution/mean": 0.7703125,
|
|
"rewards/reward_execution/std": 0.37561424523591996,
|
|
"rewards/reward_format/mean": 0.041875000577419996,
|
|
"rewards/reward_format/std": 0.1318213701248169,
|
|
"rewards/reward_length/mean": -0.0010006023359892425,
|
|
"rewards/reward_length/std": 0.0035633786304970273,
|
|
"step": 60,
|
|
"step_time": 31.793317629499825
|
|
},
|
|
{
|
|
"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": 0.115625,
|
|
"completions/max_length": 63.6,
|
|
"completions/max_terminated_length": 56.9,
|
|
"completions/mean_length": 29.54375,
|
|
"completions/mean_terminated_length": 25.17157325744629,
|
|
"completions/min_length": 6.9,
|
|
"completions/min_terminated_length": 6.9,
|
|
"entropy": 0.018630545167616218,
|
|
"epoch": 0.1830065359477124,
|
|
"frac_reward_zero_std": 0.91875,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.7876923076923075e-06,
|
|
"loss": 0.0015908852219581604,
|
|
"num_tokens": 697255.0,
|
|
"reward": 0.840159124135971,
|
|
"reward_std": 0.4418466717004776,
|
|
"rewards/reward_execution/mean": 0.8078125,
|
|
"rewards/reward_execution/std": 0.3467457413673401,
|
|
"rewards/reward_format/mean": 0.032500000391155484,
|
|
"rewards/reward_format/std": 0.15182443857192993,
|
|
"rewards/reward_length/mean": -0.00015340613426815253,
|
|
"rewards/reward_length/std": 0.0007915932299511041,
|
|
"step": 70,
|
|
"step_time": 49.57826432340007
|
|
},
|
|
{
|
|
"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": 0.134375,
|
|
"completions/max_length": 60.4,
|
|
"completions/max_terminated_length": 54.1,
|
|
"completions/mean_length": 33.16875,
|
|
"completions/mean_terminated_length": 28.681942176818847,
|
|
"completions/min_length": 7.0,
|
|
"completions/min_terminated_length": 7.0,
|
|
"entropy": 0.01808294388829381,
|
|
"epoch": 0.20915032679738563,
|
|
"frac_reward_zero_std": 0.8625,
|
|
"grad_norm": 0.15154701471328735,
|
|
"learning_rate": 1.756923076923077e-06,
|
|
"loss": -0.002125771716237068,
|
|
"num_tokens": 817325.0,
|
|
"reward": 0.7154848396778106,
|
|
"reward_std": 0.5426658689975739,
|
|
"rewards/reward_execution/mean": 0.7328125,
|
|
"rewards/reward_execution/std": 0.4067409634590149,
|
|
"rewards/reward_format/mean": -0.016250000242143868,
|
|
"rewards/reward_format/std": 0.22525277733802795,
|
|
"rewards/reward_length/mean": -0.0010776955514302244,
|
|
"rewards/reward_length/std": 0.0030558754071535075,
|
|
"step": 80,
|
|
"step_time": 40.666842412899996
|
|
},
|
|
{
|
|
"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": 0.13125,
|
|
"completions/max_length": 62.0,
|
|
"completions/max_terminated_length": 57.4,
|
|
"completions/mean_length": 30.84375,
|
|
"completions/mean_terminated_length": 25.944643592834474,
|
|
"completions/min_length": 4.4,
|
|
"completions/min_terminated_length": 4.4,
|
|
"entropy": 0.017996305134784053,
|
|
"epoch": 0.23529411764705882,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.7261538461538461e-06,
|
|
"loss": 0.0011274717748165131,
|
|
"num_tokens": 917615.0,
|
|
"reward": 0.80816171169281,
|
|
"reward_std": 0.44731455147266386,
|
|
"rewards/reward_execution/mean": 0.80625,
|
|
"rewards/reward_execution/std": 0.3403610408306122,
|
|
"rewards/reward_format/mean": 0.00250000087544322,
|
|
"rewards/reward_format/std": 0.17354516237974166,
|
|
"rewards/reward_length/mean": -0.0005883172758331057,
|
|
"rewards/reward_length/std": 0.002056805709071341,
|
|
"step": 90,
|
|
"step_time": 31.685842761999858
|
|
},
|
|
{
|
|
"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": 0.0625,
|
|
"completions/max_length": 61.5,
|
|
"completions/max_terminated_length": 57.8,
|
|
"completions/mean_length": 30.646875,
|
|
"completions/mean_terminated_length": 28.49464359283447,
|
|
"completions/min_length": 5.8,
|
|
"completions/min_terminated_length": 5.8,
|
|
"entropy": 0.013430354847150739,
|
|
"epoch": 0.26143790849673204,
|
|
"frac_reward_zero_std": 0.94375,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.6953846153846153e-06,
|
|
"loss": -0.001813487708568573,
|
|
"num_tokens": 1006586.0,
|
|
"reward": 0.8612240970134735,
|
|
"reward_std": 0.35794743039878085,
|
|
"rewards/reward_execution/mean": 0.84375,
|
|
"rewards/reward_execution/std": 0.2848026275634766,
|
|
"rewards/reward_format/mean": 0.017500000447034834,
|
|
"rewards/reward_format/std": 0.14063476622104645,
|
|
"rewards/reward_length/mean": -2.5934219411283266e-05,
|
|
"rewards/reward_length/std": 0.0001467060876166215,
|
|
"step": 100,
|
|
"step_time": 21.489991974300118
|
|
},
|
|
{
|
|
"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": 0.071875,
|
|
"completions/max_length": 57.3,
|
|
"completions/max_terminated_length": 49.5,
|
|
"completions/mean_length": 29.984375,
|
|
"completions/mean_terminated_length": 27.33921871185303,
|
|
"completions/min_length": 10.2,
|
|
"completions/min_terminated_length": 10.2,
|
|
"entropy": 0.020688600114408474,
|
|
"epoch": 0.2875816993464052,
|
|
"frac_reward_zero_std": 0.90625,
|
|
"grad_norm": 0.2687777280807495,
|
|
"learning_rate": 1.6646153846153845e-06,
|
|
"loss": 0.0025101430714130402,
|
|
"num_tokens": 1102553.0,
|
|
"reward": 0.8811502993106842,
|
|
"reward_std": 0.41800577044487,
|
|
"rewards/reward_execution/mean": 0.834375,
|
|
"rewards/reward_execution/std": 0.3469096183776855,
|
|
"rewards/reward_format/mean": 0.047500000707805155,
|
|
"rewards/reward_format/std": 0.10558144599199296,
|
|
"rewards/reward_length/mean": -0.0007247258254210465,
|
|
"rewards/reward_length/std": 0.0028026591240632114,
|
|
"step": 110,
|
|
"step_time": 23.156134998399057
|
|
},
|
|
{
|
|
"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": 0.078125,
|
|
"completions/max_length": 62.3,
|
|
"completions/max_terminated_length": 58.2,
|
|
"completions/mean_length": 32.528125,
|
|
"completions/mean_terminated_length": 29.87061176300049,
|
|
"completions/min_length": 5.0,
|
|
"completions/min_terminated_length": 5.0,
|
|
"entropy": 0.02554786999571661,
|
|
"epoch": 0.3137254901960784,
|
|
"frac_reward_zero_std": 0.88125,
|
|
"grad_norm": 0.18292272090911865,
|
|
"learning_rate": 1.6338461538461539e-06,
|
|
"loss": 0.0028493616729974745,
|
|
"num_tokens": 1206938.0,
|
|
"reward": 0.8099652647972106,
|
|
"reward_std": 0.46784267723560335,
|
|
"rewards/reward_execution/mean": 0.8,
|
|
"rewards/reward_execution/std": 0.3461746141314507,
|
|
"rewards/reward_format/mean": 0.009999999590218068,
|
|
"rewards/reward_format/std": 0.20636706799268723,
|
|
"rewards/reward_length/mean": -3.4767288889270274e-05,
|
|
"rewards/reward_length/std": 0.00017092272973968649,
|
|
"step": 120,
|
|
"step_time": 33.13626013119974
|
|
},
|
|
{
|
|
"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": 0.06875,
|
|
"completions/max_length": 62.4,
|
|
"completions/max_terminated_length": 59.2,
|
|
"completions/mean_length": 29.51875,
|
|
"completions/mean_terminated_length": 26.827375411987305,
|
|
"completions/min_length": 7.2,
|
|
"completions/min_terminated_length": 7.2,
|
|
"entropy": 0.021981789339588432,
|
|
"epoch": 0.33986928104575165,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 0.17666597664356232,
|
|
"learning_rate": 1.603076923076923e-06,
|
|
"loss": -0.007416331022977829,
|
|
"num_tokens": 1310532.0,
|
|
"reward": 0.8539650678634644,
|
|
"reward_std": 0.4202278614044189,
|
|
"rewards/reward_execution/mean": 0.825,
|
|
"rewards/reward_execution/std": 0.33841515332460403,
|
|
"rewards/reward_format/mean": 0.0306250000372529,
|
|
"rewards/reward_format/std": 0.1260850578546524,
|
|
"rewards/reward_length/mean": -0.0016599697235506028,
|
|
"rewards/reward_length/std": 0.004862861796209472,
|
|
"step": 130,
|
|
"step_time": 32.03501447449962
|
|
},
|
|
{
|
|
"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": 0.04375,
|
|
"completions/max_length": 56.9,
|
|
"completions/max_terminated_length": 52.9,
|
|
"completions/mean_length": 24.625,
|
|
"completions/mean_terminated_length": 22.959626579284667,
|
|
"completions/min_length": 1.6,
|
|
"completions/min_terminated_length": 1.6,
|
|
"entropy": 0.011234006816266628,
|
|
"epoch": 0.3660130718954248,
|
|
"frac_reward_zero_std": 0.9375,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.5723076923076923e-06,
|
|
"loss": -0.00020511075854301453,
|
|
"num_tokens": 1399128.0,
|
|
"reward": 0.8977922797203064,
|
|
"reward_std": 0.3633412554860115,
|
|
"rewards/reward_execution/mean": 0.903125,
|
|
"rewards/reward_execution/std": 0.2428273230791092,
|
|
"rewards/reward_format/mean": -0.004999999888241291,
|
|
"rewards/reward_format/std": 0.20932087153196335,
|
|
"rewards/reward_length/mean": -0.0003327726615680149,
|
|
"rewards/reward_length/std": 0.0015821949622477406,
|
|
"step": 140,
|
|
"step_time": 21.063309571300124
|
|
},
|
|
{
|
|
"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": 0.096875,
|
|
"completions/max_length": 60.2,
|
|
"completions/max_terminated_length": 50.5,
|
|
"completions/mean_length": 28.821875,
|
|
"completions/mean_terminated_length": 25.15128993988037,
|
|
"completions/min_length": 7.2,
|
|
"completions/min_terminated_length": 7.2,
|
|
"entropy": 0.01741225158511952,
|
|
"epoch": 0.39215686274509803,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.2630400061607361,
|
|
"learning_rate": 1.5415384615384614e-06,
|
|
"loss": 0.0048006601631641384,
|
|
"num_tokens": 1488523.0,
|
|
"reward": 0.8722868859767914,
|
|
"reward_std": 0.40390884727239607,
|
|
"rewards/reward_execution/mean": 0.8375,
|
|
"rewards/reward_execution/std": 0.3128639951348305,
|
|
"rewards/reward_format/mean": 0.03624999956227839,
|
|
"rewards/reward_format/std": 0.1403054416179657,
|
|
"rewards/reward_length/mean": -0.001463154905650299,
|
|
"rewards/reward_length/std": 0.004689511815013247,
|
|
"step": 150,
|
|
"step_time": 29.140895940899373
|
|
},
|
|
{
|
|
"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": 0.1,
|
|
"completions/max_length": 64.0,
|
|
"completions/max_terminated_length": 52.8,
|
|
"completions/mean_length": 29.028125,
|
|
"completions/mean_terminated_length": 25.344496536254884,
|
|
"completions/min_length": 5.0,
|
|
"completions/min_terminated_length": 5.0,
|
|
"entropy": 0.014457397866863175,
|
|
"epoch": 0.41830065359477125,
|
|
"frac_reward_zero_std": 0.90625,
|
|
"grad_norm": 0.24363495409488678,
|
|
"learning_rate": 1.5107692307692308e-06,
|
|
"loss": 3.446638584136963e-06,
|
|
"num_tokens": 1597840.0,
|
|
"reward": 0.7838960945606231,
|
|
"reward_std": 0.5096011608839035,
|
|
"rewards/reward_execution/mean": 0.8046875,
|
|
"rewards/reward_execution/std": 0.38871022760868074,
|
|
"rewards/reward_format/mean": -0.02000000039115548,
|
|
"rewards/reward_format/std": 0.2278716519474983,
|
|
"rewards/reward_length/mean": -0.0007914238143712282,
|
|
"rewards/reward_length/std": 0.0027368536102585495,
|
|
"step": 160,
|
|
"step_time": 33.41009088679984
|
|
},
|
|
{
|
|
"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": 0.13125,
|
|
"completions/max_length": 58.7,
|
|
"completions/max_terminated_length": 48.9,
|
|
"completions/mean_length": 31.903125,
|
|
"completions/mean_terminated_length": 26.545238304138184,
|
|
"completions/min_length": 6.3,
|
|
"completions/min_terminated_length": 6.3,
|
|
"entropy": 0.01579824187028862,
|
|
"epoch": 0.4444444444444444,
|
|
"frac_reward_zero_std": 0.89375,
|
|
"grad_norm": 0.3457740545272827,
|
|
"learning_rate": 1.48e-06,
|
|
"loss": -0.0016400977969169618,
|
|
"num_tokens": 1696197.0,
|
|
"reward": 0.8029215812683106,
|
|
"reward_std": 0.5009407073259353,
|
|
"rewards/reward_execution/mean": 0.815625,
|
|
"rewards/reward_execution/std": 0.36931134164333346,
|
|
"rewards/reward_format/mean": -0.012499999394640327,
|
|
"rewards/reward_format/std": 0.21184270530939103,
|
|
"rewards/reward_length/mean": -0.00020343956566648558,
|
|
"rewards/reward_length/std": 0.0009343125333543867,
|
|
"step": 170,
|
|
"step_time": 35.535485880599666
|
|
},
|
|
{
|
|
"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": 0.1125,
|
|
"completions/max_length": 60.8,
|
|
"completions/max_terminated_length": 54.0,
|
|
"completions/mean_length": 28.421875,
|
|
"completions/mean_terminated_length": 23.877572059631348,
|
|
"completions/min_length": 4.5,
|
|
"completions/min_terminated_length": 4.5,
|
|
"entropy": 0.026755200948537096,
|
|
"epoch": 0.47058823529411764,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.2557838261127472,
|
|
"learning_rate": 1.4492307692307692e-06,
|
|
"loss": 0.003511914610862732,
|
|
"num_tokens": 1811596.0,
|
|
"reward": 0.7582438230514527,
|
|
"reward_std": 0.4795005351305008,
|
|
"rewards/reward_execution/mean": 0.7484375,
|
|
"rewards/reward_execution/std": 0.40075716376304626,
|
|
"rewards/reward_format/mean": 0.010000000149011612,
|
|
"rewards/reward_format/std": 0.17314954847097397,
|
|
"rewards/reward_length/mean": -0.00019370392201381036,
|
|
"rewards/reward_length/std": 0.0010883385682973312,
|
|
"step": 180,
|
|
"step_time": 35.30616569360027
|
|
},
|
|
{
|
|
"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": 0.065625,
|
|
"completions/max_length": 61.7,
|
|
"completions/max_terminated_length": 59.7,
|
|
"completions/mean_length": 28.978125,
|
|
"completions/mean_terminated_length": 26.54868984222412,
|
|
"completions/min_length": 4.1,
|
|
"completions/min_terminated_length": 4.1,
|
|
"entropy": 0.01608970392917399,
|
|
"epoch": 0.49673202614379086,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.10139451175928116,
|
|
"learning_rate": 1.4184615384615384e-06,
|
|
"loss": 0.0025175046175718307,
|
|
"num_tokens": 1908949.0,
|
|
"reward": 0.8265464305877686,
|
|
"reward_std": 0.3705518566071987,
|
|
"rewards/reward_execution/mean": 0.8109375,
|
|
"rewards/reward_execution/std": 0.3359261803328991,
|
|
"rewards/reward_format/mean": 0.015625000419095158,
|
|
"rewards/reward_format/std": 0.16492117643356324,
|
|
"rewards/reward_length/mean": -1.60989337018691e-05,
|
|
"rewards/reward_length/std": 7.88968307460891e-05,
|
|
"step": 190,
|
|
"step_time": 30.834952960299596
|
|
},
|
|
{
|
|
"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": 0.09375,
|
|
"completions/max_length": 57.4,
|
|
"completions/max_terminated_length": 52.4,
|
|
"completions/mean_length": 27.65,
|
|
"completions/mean_terminated_length": 23.781375885009766,
|
|
"completions/min_length": 2.6,
|
|
"completions/min_terminated_length": 2.6,
|
|
"entropy": 0.014648127416057833,
|
|
"epoch": 0.5228758169934641,
|
|
"frac_reward_zero_std": 0.925,
|
|
"grad_norm": 0.11480975896120071,
|
|
"learning_rate": 1.3876923076923078e-06,
|
|
"loss": -0.005061210319399834,
|
|
"num_tokens": 2003989.0,
|
|
"reward": 0.7860415637493133,
|
|
"reward_std": 0.42728935182094574,
|
|
"rewards/reward_execution/mean": 0.80625,
|
|
"rewards/reward_execution/std": 0.3499292016029358,
|
|
"rewards/reward_format/mean": -0.02000000011175871,
|
|
"rewards/reward_format/std": 0.22197649329900743,
|
|
"rewards/reward_length/mean": -0.00020847387204412371,
|
|
"rewards/reward_length/std": 0.0011036201613023877,
|
|
"step": 200,
|
|
"step_time": 22.863012499799062
|
|
},
|
|
{
|
|
"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": 0.096875,
|
|
"completions/max_length": 54.9,
|
|
"completions/max_terminated_length": 46.0,
|
|
"completions/mean_length": 26.328125,
|
|
"completions/mean_terminated_length": 22.390333557128905,
|
|
"completions/min_length": 4.3,
|
|
"completions/min_terminated_length": 4.3,
|
|
"entropy": 0.012826071881136158,
|
|
"epoch": 0.5490196078431373,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.24089939892292023,
|
|
"learning_rate": 1.3569230769230768e-06,
|
|
"loss": -0.00509924441576004,
|
|
"num_tokens": 2088394.0,
|
|
"reward": 0.8083807528018951,
|
|
"reward_std": 0.46078217923641207,
|
|
"rewards/reward_execution/mean": 0.834375,
|
|
"rewards/reward_execution/std": 0.3126445934176445,
|
|
"rewards/reward_format/mean": -0.025624999962747096,
|
|
"rewards/reward_format/std": 0.22389649897813796,
|
|
"rewards/reward_length/mean": -0.00036928153422195465,
|
|
"rewards/reward_length/std": 0.0012297062785364687,
|
|
"step": 210,
|
|
"step_time": 19.783136276900677
|
|
},
|
|
{
|
|
"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": 0.128125,
|
|
"completions/max_length": 62.7,
|
|
"completions/max_terminated_length": 54.2,
|
|
"completions/mean_length": 30.45625,
|
|
"completions/mean_terminated_length": 25.517249488830565,
|
|
"completions/min_length": 8.1,
|
|
"completions/min_terminated_length": 8.1,
|
|
"entropy": 0.02007206030521047,
|
|
"epoch": 0.5751633986928104,
|
|
"frac_reward_zero_std": 0.925,
|
|
"grad_norm": 0.4141358435153961,
|
|
"learning_rate": 1.3261538461538461e-06,
|
|
"loss": -0.00032107755541801455,
|
|
"num_tokens": 2193168.0,
|
|
"reward": 0.8757237374782563,
|
|
"reward_std": 0.4225275322794914,
|
|
"rewards/reward_execution/mean": 0.8359375,
|
|
"rewards/reward_execution/std": 0.3276398167014122,
|
|
"rewards/reward_format/mean": 0.04000000040978193,
|
|
"rewards/reward_format/std": 0.1567118600010872,
|
|
"rewards/reward_length/mean": -0.0002137999312253669,
|
|
"rewards/reward_length/std": 0.0009932070854119957,
|
|
"step": 220,
|
|
"step_time": 24.101320211100482
|
|
},
|
|
{
|
|
"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": 0.0875,
|
|
"completions/max_length": 60.2,
|
|
"completions/max_terminated_length": 55.4,
|
|
"completions/mean_length": 27.603125,
|
|
"completions/mean_terminated_length": 24.094705295562743,
|
|
"completions/min_length": 4.0,
|
|
"completions/min_terminated_length": 4.0,
|
|
"entropy": 0.015776164229464484,
|
|
"epoch": 0.6013071895424836,
|
|
"frac_reward_zero_std": 0.94375,
|
|
"grad_norm": 0.04463358595967293,
|
|
"learning_rate": 1.2953846153846153e-06,
|
|
"loss": 0.0019590660929679872,
|
|
"num_tokens": 2290713.0,
|
|
"reward": 0.8192518889904022,
|
|
"reward_std": 0.43653333485126494,
|
|
"rewards/reward_execution/mean": 0.846875,
|
|
"rewards/reward_execution/std": 0.3155195638537407,
|
|
"rewards/reward_format/mean": -0.027500000223517417,
|
|
"rewards/reward_format/std": 0.22821232974529265,
|
|
"rewards/reward_length/mean": -0.000123155934670649,
|
|
"rewards/reward_length/std": 0.000696675164726912,
|
|
"step": 230,
|
|
"step_time": 22.918984739600727
|
|
},
|
|
{
|
|
"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": 0.11875,
|
|
"completions/max_length": 56.1,
|
|
"completions/max_terminated_length": 52.3,
|
|
"completions/mean_length": 29.221875,
|
|
"completions/mean_terminated_length": 24.746478462219237,
|
|
"completions/min_length": 5.7,
|
|
"completions/min_terminated_length": 5.7,
|
|
"entropy": 0.022762688572038316,
|
|
"epoch": 0.6274509803921569,
|
|
"frac_reward_zero_std": 0.925,
|
|
"grad_norm": 0.1914663314819336,
|
|
"learning_rate": 1.2646153846153847e-06,
|
|
"loss": 0.0010617958381772042,
|
|
"num_tokens": 2382756.0,
|
|
"reward": 0.86337229013443,
|
|
"reward_std": 0.36463922560214995,
|
|
"rewards/reward_execution/mean": 0.828125,
|
|
"rewards/reward_execution/std": 0.31460989713668824,
|
|
"rewards/reward_format/mean": 0.03624999988824129,
|
|
"rewards/reward_format/std": 0.1411946162581444,
|
|
"rewards/reward_length/mean": -0.001002733207133133,
|
|
"rewards/reward_length/std": 0.0036312009207904337,
|
|
"step": 240,
|
|
"step_time": 21.189084481199096
|
|
},
|
|
{
|
|
"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": 0.08125,
|
|
"completions/max_length": 58.2,
|
|
"completions/max_terminated_length": 54.0,
|
|
"completions/mean_length": 28.728125,
|
|
"completions/mean_terminated_length": 25.495055198669434,
|
|
"completions/min_length": 4.8,
|
|
"completions/min_terminated_length": 4.8,
|
|
"entropy": 0.014168781463376945,
|
|
"epoch": 0.6535947712418301,
|
|
"frac_reward_zero_std": 0.9375,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.2338461538461537e-06,
|
|
"loss": 0.002245112136006355,
|
|
"num_tokens": 2501901.0,
|
|
"reward": 0.8953510224819183,
|
|
"reward_std": 0.36954851299524305,
|
|
"rewards/reward_execution/mean": 0.859375,
|
|
"rewards/reward_execution/std": 0.28630427122116087,
|
|
"rewards/reward_format/mean": 0.03625000072643161,
|
|
"rewards/reward_format/std": 0.13394820243120192,
|
|
"rewards/reward_length/mean": -0.0002740118248766521,
|
|
"rewards/reward_length/std": 0.001044175514834933,
|
|
"step": 250,
|
|
"step_time": 47.213486583600385
|
|
},
|
|
{
|
|
"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": 0.084375,
|
|
"completions/max_length": 63.1,
|
|
"completions/max_terminated_length": 54.7,
|
|
"completions/mean_length": 28.525,
|
|
"completions/mean_terminated_length": 24.897054862976074,
|
|
"completions/min_length": 4.3,
|
|
"completions/min_terminated_length": 4.3,
|
|
"entropy": 0.015173432582378155,
|
|
"epoch": 0.6797385620915033,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 0.3051759600639343,
|
|
"learning_rate": 1.203076923076923e-06,
|
|
"loss": 0.007945796847343445,
|
|
"num_tokens": 2595077.0,
|
|
"reward": 0.8287680983543396,
|
|
"reward_std": 0.4496727645397186,
|
|
"rewards/reward_execution/mean": 0.8546875,
|
|
"rewards/reward_execution/std": 0.33965854346752167,
|
|
"rewards/reward_format/mean": -0.025624999683350323,
|
|
"rewards/reward_format/std": 0.22063558399677277,
|
|
"rewards/reward_length/mean": -0.00029443361345329324,
|
|
"rewards/reward_length/std": 0.0014556680456735195,
|
|
"step": 260,
|
|
"step_time": 22.569890121100798
|
|
},
|
|
{
|
|
"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": 0.0875,
|
|
"completions/max_length": 58.0,
|
|
"completions/max_terminated_length": 51.9,
|
|
"completions/mean_length": 28.290625,
|
|
"completions/mean_terminated_length": 24.7339298248291,
|
|
"completions/min_length": 4.0,
|
|
"completions/min_terminated_length": 4.0,
|
|
"entropy": 0.01848839840185974,
|
|
"epoch": 0.7058823529411765,
|
|
"frac_reward_zero_std": 0.88125,
|
|
"grad_norm": 0.16644737124443054,
|
|
"learning_rate": 1.1723076923076923e-06,
|
|
"loss": 0.007904282957315444,
|
|
"num_tokens": 2688022.0,
|
|
"reward": 0.8564159631729126,
|
|
"reward_std": 0.38055695593357086,
|
|
"rewards/reward_execution/mean": 0.840625,
|
|
"rewards/reward_execution/std": 0.32737596333026886,
|
|
"rewards/reward_format/mean": 0.017500000633299352,
|
|
"rewards/reward_format/std": 0.15969619601964952,
|
|
"rewards/reward_length/mean": -0.0017090746201574802,
|
|
"rewards/reward_length/std": 0.005349089484661817,
|
|
"step": 270,
|
|
"step_time": 25.10648124489908
|
|
},
|
|
{
|
|
"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": 0.090625,
|
|
"completions/max_length": 61.8,
|
|
"completions/max_terminated_length": 53.5,
|
|
"completions/mean_length": 25.365625,
|
|
"completions/mean_terminated_length": 21.186185836791992,
|
|
"completions/min_length": 4.3,
|
|
"completions/min_terminated_length": 4.3,
|
|
"entropy": 0.016150319923508506,
|
|
"epoch": 0.7320261437908496,
|
|
"frac_reward_zero_std": 0.91875,
|
|
"grad_norm": 0.262028306722641,
|
|
"learning_rate": 1.1415384615384615e-06,
|
|
"loss": 0.006609904766082764,
|
|
"num_tokens": 2789795.0,
|
|
"reward": 0.7981519639492035,
|
|
"reward_std": 0.4216637030243874,
|
|
"rewards/reward_execution/mean": 0.8296875,
|
|
"rewards/reward_execution/std": 0.31778890937566756,
|
|
"rewards/reward_format/mean": -0.031250000558793546,
|
|
"rewards/reward_format/std": 0.20076166242361068,
|
|
"rewards/reward_length/mean": -0.0002855546917999163,
|
|
"rewards/reward_length/std": 0.0014620612091675866,
|
|
"step": 280,
|
|
"step_time": 37.48035952199989
|
|
},
|
|
{
|
|
"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": 0.034375,
|
|
"completions/max_length": 56.4,
|
|
"completions/max_terminated_length": 55.1,
|
|
"completions/mean_length": 26.109375,
|
|
"completions/mean_terminated_length": 24.60078582763672,
|
|
"completions/min_length": 4.6,
|
|
"completions/min_terminated_length": 4.6,
|
|
"entropy": 0.013258904079884814,
|
|
"epoch": 0.7581699346405228,
|
|
"frac_reward_zero_std": 0.925,
|
|
"grad_norm": 0.1547228842973709,
|
|
"learning_rate": 1.1107692307692306e-06,
|
|
"loss": -0.005408792942762375,
|
|
"num_tokens": 2892818.0,
|
|
"reward": 0.9263898611068726,
|
|
"reward_std": 0.33983229845762253,
|
|
"rewards/reward_execution/mean": 0.8921875,
|
|
"rewards/reward_execution/std": 0.27078352123498917,
|
|
"rewards/reward_format/mean": 0.03437500074505806,
|
|
"rewards/reward_format/std": 0.13535554111003875,
|
|
"rewards/reward_length/mean": -0.00017268856718146707,
|
|
"rewards/reward_length/std": 0.0009681309235020308,
|
|
"step": 290,
|
|
"step_time": 29.690082654099022
|
|
},
|
|
{
|
|
"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": 0.025,
|
|
"completions/max_length": 54.5,
|
|
"completions/max_terminated_length": 52.5,
|
|
"completions/mean_length": 22.4875,
|
|
"completions/mean_terminated_length": 21.38839292526245,
|
|
"completions/min_length": 2.9,
|
|
"completions/min_terminated_length": 2.9,
|
|
"entropy": 0.010033508395463287,
|
|
"epoch": 0.7843137254901961,
|
|
"frac_reward_zero_std": 0.95,
|
|
"grad_norm": 0.13593685626983643,
|
|
"learning_rate": 1.08e-06,
|
|
"loss": 0.0032884977757930756,
|
|
"num_tokens": 2986654.0,
|
|
"reward": 0.8979483902454376,
|
|
"reward_std": 0.34552761912345886,
|
|
"rewards/reward_execution/mean": 0.903125,
|
|
"rewards/reward_execution/std": 0.21679688692092897,
|
|
"rewards/reward_format/mean": -0.005000000074505806,
|
|
"rewards/reward_format/std": 0.1903120458126068,
|
|
"rewards/reward_length/mean": -0.00017664464539848267,
|
|
"rewards/reward_length/std": 0.0009606391424313188,
|
|
"step": 300,
|
|
"step_time": 34.64459892829982
|
|
},
|
|
{
|
|
"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": 0.025,
|
|
"completions/max_length": 54.1,
|
|
"completions/max_terminated_length": 52.9,
|
|
"completions/mean_length": 25.453125,
|
|
"completions/mean_terminated_length": 24.464732360839843,
|
|
"completions/min_length": 2.3,
|
|
"completions/min_terminated_length": 2.3,
|
|
"entropy": 0.014458585100510391,
|
|
"epoch": 0.8104575163398693,
|
|
"frac_reward_zero_std": 0.94375,
|
|
"grad_norm": 0.2604885995388031,
|
|
"learning_rate": 1.0492307692307692e-06,
|
|
"loss": 0.00032596662640571594,
|
|
"num_tokens": 3075579.0,
|
|
"reward": 0.9042167961597443,
|
|
"reward_std": 0.3212708577513695,
|
|
"rewards/reward_execution/mean": 0.909375,
|
|
"rewards/reward_execution/std": 0.18733828961849214,
|
|
"rewards/reward_format/mean": -0.004999999888241291,
|
|
"rewards/reward_format/std": 0.18719189018011093,
|
|
"rewards/reward_length/mean": -0.00015824955080461222,
|
|
"rewards/reward_length/std": 0.0008951946147135459,
|
|
"step": 310,
|
|
"step_time": 21.03584319659931
|
|
},
|
|
{
|
|
"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": 0.096875,
|
|
"completions/max_length": 63.4,
|
|
"completions/max_terminated_length": 54.5,
|
|
"completions/mean_length": 29.725,
|
|
"completions/mean_terminated_length": 26.006825065612794,
|
|
"completions/min_length": 6.1,
|
|
"completions/min_terminated_length": 6.1,
|
|
"entropy": 0.012181862572106184,
|
|
"epoch": 0.8366013071895425,
|
|
"frac_reward_zero_std": 0.95,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.0184615384615384e-06,
|
|
"loss": 0.0007298282347619533,
|
|
"num_tokens": 3167899.0,
|
|
"reward": 0.8702065765857696,
|
|
"reward_std": 0.4533663779497147,
|
|
"rewards/reward_execution/mean": 0.85,
|
|
"rewards/reward_execution/std": 0.3251454561948776,
|
|
"rewards/reward_format/mean": 0.021249999664723872,
|
|
"rewards/reward_format/std": 0.18280272781848908,
|
|
"rewards/reward_length/mean": -0.0010434576674015262,
|
|
"rewards/reward_length/std": 0.0033523349375172984,
|
|
"step": 320,
|
|
"step_time": 23.372393020300116
|
|
},
|
|
{
|
|
"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": 0.096875,
|
|
"completions/max_length": 56.0,
|
|
"completions/max_terminated_length": 52.2,
|
|
"completions/mean_length": 26.428125,
|
|
"completions/mean_terminated_length": 22.613398361206055,
|
|
"completions/min_length": 5.9,
|
|
"completions/min_terminated_length": 5.9,
|
|
"entropy": 0.02121344533697993,
|
|
"epoch": 0.8627450980392157,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.1595773994922638,
|
|
"learning_rate": 9.876923076923076e-07,
|
|
"loss": 0.004875291138887405,
|
|
"num_tokens": 3274320.0,
|
|
"reward": 0.8352176487445832,
|
|
"reward_std": 0.43423864543437957,
|
|
"rewards/reward_execution/mean": 0.8234375,
|
|
"rewards/reward_execution/std": 0.3197143077850342,
|
|
"rewards/reward_format/mean": 0.011874999292194844,
|
|
"rewards/reward_format/std": 0.1914305344223976,
|
|
"rewards/reward_length/mean": -9.489577569183893e-05,
|
|
"rewards/reward_length/std": 0.00042889252363238485,
|
|
"step": 330,
|
|
"step_time": 38.35784068989906
|
|
},
|
|
{
|
|
"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": 0.06875,
|
|
"completions/max_length": 57.8,
|
|
"completions/max_terminated_length": 54.6,
|
|
"completions/mean_length": 27.5625,
|
|
"completions/mean_terminated_length": 24.792351913452148,
|
|
"completions/min_length": 3.9,
|
|
"completions/min_terminated_length": 3.9,
|
|
"entropy": 0.02091266990046279,
|
|
"epoch": 0.8888888888888888,
|
|
"frac_reward_zero_std": 0.90625,
|
|
"grad_norm": 0.0920712947845459,
|
|
"learning_rate": 9.569230769230768e-07,
|
|
"loss": 5.366690456867218e-05,
|
|
"num_tokens": 3369568.0,
|
|
"reward": 0.8437139749526977,
|
|
"reward_std": 0.39172282814979553,
|
|
"rewards/reward_execution/mean": 0.8453125,
|
|
"rewards/reward_execution/std": 0.29225673228502275,
|
|
"rewards/reward_format/mean": -0.001249999739229679,
|
|
"rewards/reward_format/std": 0.1931886628270149,
|
|
"rewards/reward_length/mean": -0.0003485622819425771,
|
|
"rewards/reward_length/std": 0.0019275649421615527,
|
|
"step": 340,
|
|
"step_time": 30.992745630099307
|
|
},
|
|
{
|
|
"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": 0.115625,
|
|
"completions/max_length": 63.5,
|
|
"completions/max_terminated_length": 54.1,
|
|
"completions/mean_length": 30.60625,
|
|
"completions/mean_terminated_length": 26.010983848571776,
|
|
"completions/min_length": 5.2,
|
|
"completions/min_terminated_length": 5.2,
|
|
"entropy": 0.013417703769482614,
|
|
"epoch": 0.9150326797385621,
|
|
"frac_reward_zero_std": 0.90625,
|
|
"grad_norm": 0.37030595541000366,
|
|
"learning_rate": 9.261538461538461e-07,
|
|
"loss": -0.0009939376264810561,
|
|
"num_tokens": 3464962.0,
|
|
"reward": 0.8491355240345001,
|
|
"reward_std": 0.4328867569565773,
|
|
"rewards/reward_execution/mean": 0.84375,
|
|
"rewards/reward_execution/std": 0.33426086604595184,
|
|
"rewards/reward_format/mean": 0.006249999813735485,
|
|
"rewards/reward_format/std": 0.18620961904525757,
|
|
"rewards/reward_length/mean": -0.0008644971967441961,
|
|
"rewards/reward_length/std": 0.002869903548707953,
|
|
"step": 350,
|
|
"step_time": 30.502866643200104
|
|
},
|
|
{
|
|
"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": 0.078125,
|
|
"completions/max_length": 59.6,
|
|
"completions/max_terminated_length": 56.8,
|
|
"completions/mean_length": 28.43125,
|
|
"completions/mean_terminated_length": 25.694689750671387,
|
|
"completions/min_length": 6.6,
|
|
"completions/min_terminated_length": 6.6,
|
|
"entropy": 0.017781620883761208,
|
|
"epoch": 0.9411764705882353,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 8.953846153846152e-07,
|
|
"loss": 0.0047313995659351345,
|
|
"num_tokens": 3564220.0,
|
|
"reward": 0.8438348054885865,
|
|
"reward_std": 0.37721367329359057,
|
|
"rewards/reward_execution/mean": 0.821875,
|
|
"rewards/reward_execution/std": 0.33306297212839125,
|
|
"rewards/reward_format/mean": 0.023124999599531293,
|
|
"rewards/reward_format/std": 0.1382562428712845,
|
|
"rewards/reward_length/mean": -0.0011652146029518917,
|
|
"rewards/reward_length/std": 0.0036097699485253543,
|
|
"step": 360,
|
|
"step_time": 23.791317949199584
|
|
},
|
|
{
|
|
"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": 0.171875,
|
|
"completions/max_length": 62.0,
|
|
"completions/max_terminated_length": 49.2,
|
|
"completions/mean_length": 30.690625,
|
|
"completions/mean_terminated_length": 24.01605615615845,
|
|
"completions/min_length": 5.9,
|
|
"completions/min_terminated_length": 5.9,
|
|
"entropy": 0.021915445105059918,
|
|
"epoch": 0.9673202614379085,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.3232307434082031,
|
|
"learning_rate": 8.646153846153845e-07,
|
|
"loss": 0.0052782092243433,
|
|
"num_tokens": 3665201.0,
|
|
"reward": 0.774853527545929,
|
|
"reward_std": 0.49129743278026583,
|
|
"rewards/reward_execution/mean": 0.775,
|
|
"rewards/reward_execution/std": 0.35995189696550367,
|
|
"rewards/reward_format/mean": 0.0006250000093132258,
|
|
"rewards/reward_format/std": 0.2222874268889427,
|
|
"rewards/reward_length/mean": -0.0007715050311162486,
|
|
"rewards/reward_length/std": 0.0025452338199102088,
|
|
"step": 370,
|
|
"step_time": 29.337286553499872
|
|
},
|
|
{
|
|
"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": 0.078125,
|
|
"completions/max_length": 61.3,
|
|
"completions/max_terminated_length": 57.7,
|
|
"completions/mean_length": 29.89375,
|
|
"completions/mean_terminated_length": 27.315965843200683,
|
|
"completions/min_length": 6.4,
|
|
"completions/min_terminated_length": 6.4,
|
|
"entropy": 0.01449486033507128,
|
|
"epoch": 0.9934640522875817,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 8.338461538461537e-07,
|
|
"loss": 0.0016757490113377572,
|
|
"num_tokens": 3774267.0,
|
|
"reward": 0.8673543214797974,
|
|
"reward_std": 0.3941518574953079,
|
|
"rewards/reward_execution/mean": 0.8484375,
|
|
"rewards/reward_execution/std": 0.2857915133237839,
|
|
"rewards/reward_format/mean": 0.01937499991618097,
|
|
"rewards/reward_format/std": 0.17438557744026184,
|
|
"rewards/reward_length/mean": -0.00045820135837857376,
|
|
"rewards/reward_length/std": 0.002131153648588224,
|
|
"step": 380,
|
|
"step_time": 39.54191667520117
|
|
},
|
|
{
|
|
"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": 0.08125,
|
|
"completions/max_length": 57.2,
|
|
"completions/max_terminated_length": 53.2,
|
|
"completions/mean_length": 28.64375,
|
|
"completions/mean_terminated_length": 25.722783470153807,
|
|
"completions/min_length": 4.2,
|
|
"completions/min_terminated_length": 4.2,
|
|
"entropy": 0.016135018773734612,
|
|
"epoch": 1.0183006535947712,
|
|
"frac_reward_zero_std": 0.925,
|
|
"grad_norm": 0.010175619274377823,
|
|
"learning_rate": 8.03076923076923e-07,
|
|
"loss": -0.0021699491888284685,
|
|
"num_tokens": 3864905.0,
|
|
"reward": 0.8768511593341828,
|
|
"reward_std": 0.3708808749914169,
|
|
"rewards/reward_execution/mean": 0.8671875,
|
|
"rewards/reward_execution/std": 0.2934316545724869,
|
|
"rewards/reward_format/mean": 0.010000000149011612,
|
|
"rewards/reward_format/std": 0.20015330314636232,
|
|
"rewards/reward_length/mean": -0.00033637472661212086,
|
|
"rewards/reward_length/std": 0.0017045978922396899,
|
|
"step": 390,
|
|
"step_time": 25.744064228600838
|
|
},
|
|
{
|
|
"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": 0.065625,
|
|
"completions/max_length": 62.3,
|
|
"completions/max_terminated_length": 56.5,
|
|
"completions/mean_length": 28.453125,
|
|
"completions/mean_terminated_length": 25.64851818084717,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"entropy": 0.0148160065364209,
|
|
"epoch": 1.0444444444444445,
|
|
"frac_reward_zero_std": 0.9375,
|
|
"grad_norm": 0.16944339871406555,
|
|
"learning_rate": 7.723076923076923e-07,
|
|
"loss": 0.0012604285031557084,
|
|
"num_tokens": 3958542.0,
|
|
"reward": 0.9116561233997345,
|
|
"reward_std": 0.3625919044017792,
|
|
"rewards/reward_execution/mean": 0.8625,
|
|
"rewards/reward_execution/std": 0.3148761510848999,
|
|
"rewards/reward_format/mean": 0.049375000642612576,
|
|
"rewards/reward_format/std": 0.1376934230327606,
|
|
"rewards/reward_length/mean": -0.0002189006354456069,
|
|
"rewards/reward_length/std": 0.0012352988866041414,
|
|
"step": 400,
|
|
"step_time": 30.534184857298715
|
|
},
|
|
{
|
|
"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": 0.065625,
|
|
"completions/max_length": 60.6,
|
|
"completions/max_terminated_length": 58.1,
|
|
"completions/mean_length": 28.50625,
|
|
"completions/mean_terminated_length": 26.372420883178712,
|
|
"completions/min_length": 4.8,
|
|
"completions/min_terminated_length": 4.8,
|
|
"entropy": 0.019986266074738523,
|
|
"epoch": 1.0705882352941176,
|
|
"frac_reward_zero_std": 0.8625,
|
|
"grad_norm": 0.16391511261463165,
|
|
"learning_rate": 7.415384615384616e-07,
|
|
"loss": -0.0035576265305280685,
|
|
"num_tokens": 4050648.0,
|
|
"reward": 0.8699087738990784,
|
|
"reward_std": 0.40259915590286255,
|
|
"rewards/reward_execution/mean": 0.853125,
|
|
"rewards/reward_execution/std": 0.2920844301581383,
|
|
"rewards/reward_format/mean": 0.017500000167638065,
|
|
"rewards/reward_format/std": 0.1922121524810791,
|
|
"rewards/reward_length/mean": -0.0007162786248954944,
|
|
"rewards/reward_length/std": 0.0025070570129173573,
|
|
"step": 410,
|
|
"step_time": 21.825925696900594
|
|
},
|
|
{
|
|
"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": 0.1125,
|
|
"completions/max_length": 59.7,
|
|
"completions/max_terminated_length": 56.6,
|
|
"completions/mean_length": 31.790625,
|
|
"completions/mean_terminated_length": 28.320568084716797,
|
|
"completions/min_length": 8.2,
|
|
"completions/min_terminated_length": 8.2,
|
|
"entropy": 0.018230384926482657,
|
|
"epoch": 1.096732026143791,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.20660214126110077,
|
|
"learning_rate": 7.107692307692308e-07,
|
|
"loss": -0.006729409843683243,
|
|
"num_tokens": 4157733.0,
|
|
"reward": 0.8187875032424927,
|
|
"reward_std": 0.46888020187616347,
|
|
"rewards/reward_execution/mean": 0.8203125,
|
|
"rewards/reward_execution/std": 0.3394189789891243,
|
|
"rewards/reward_format/mean": -0.0012500002514570952,
|
|
"rewards/reward_format/std": 0.20192425698041916,
|
|
"rewards/reward_length/mean": -0.00027503475212142803,
|
|
"rewards/reward_length/std": 0.0013202042420743965,
|
|
"step": 420,
|
|
"step_time": 44.198702486898945
|
|
},
|
|
{
|
|
"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": 0.06875,
|
|
"completions/max_length": 57.7,
|
|
"completions/max_terminated_length": 46.0,
|
|
"completions/mean_length": 26.6,
|
|
"completions/mean_terminated_length": 23.688428688049317,
|
|
"completions/min_length": 5.3,
|
|
"completions/min_terminated_length": 5.3,
|
|
"entropy": 0.021190898454187847,
|
|
"epoch": 1.122875816993464,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.21485479176044464,
|
|
"learning_rate": 6.800000000000001e-07,
|
|
"loss": 0.005003933981060982,
|
|
"num_tokens": 4267413.0,
|
|
"reward": 0.8358113527297973,
|
|
"reward_std": 0.38059340008185244,
|
|
"rewards/reward_execution/mean": 0.8125,
|
|
"rewards/reward_execution/std": 0.29567536115646365,
|
|
"rewards/reward_format/mean": 0.024999999813735484,
|
|
"rewards/reward_format/std": 0.16171137988567352,
|
|
"rewards/reward_length/mean": -0.0016886711468032446,
|
|
"rewards/reward_length/std": 0.004632086679703207,
|
|
"step": 430,
|
|
"step_time": 47.47027277050074
|
|
},
|
|
{
|
|
"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": 0.08125,
|
|
"completions/max_length": 58.6,
|
|
"completions/max_terminated_length": 55.1,
|
|
"completions/mean_length": 29.134375,
|
|
"completions/mean_terminated_length": 26.33336181640625,
|
|
"completions/min_length": 5.6,
|
|
"completions/min_terminated_length": 5.6,
|
|
"entropy": 0.016694543626817902,
|
|
"epoch": 1.1490196078431372,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.14693917334079742,
|
|
"learning_rate": 6.492307692307692e-07,
|
|
"loss": -0.000694776140153408,
|
|
"num_tokens": 4368252.0,
|
|
"reward": 0.8373144149780274,
|
|
"reward_std": 0.44957631528377534,
|
|
"rewards/reward_execution/mean": 0.79375,
|
|
"rewards/reward_execution/std": 0.3737864479422569,
|
|
"rewards/reward_format/mean": 0.04375000069849193,
|
|
"rewards/reward_format/std": 0.13114665895700456,
|
|
"rewards/reward_length/mean": -0.00018561571905593155,
|
|
"rewards/reward_length/std": 0.0010246489950077376,
|
|
"step": 440,
|
|
"step_time": 30.12494883259933
|
|
},
|
|
{
|
|
"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": 0.08125,
|
|
"completions/max_length": 61.8,
|
|
"completions/max_terminated_length": 55.3,
|
|
"completions/mean_length": 27.6375,
|
|
"completions/mean_terminated_length": 24.231104469299318,
|
|
"completions/min_length": 2.1,
|
|
"completions/min_terminated_length": 2.1,
|
|
"entropy": 0.015231955200306402,
|
|
"epoch": 1.1751633986928105,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.30323952436447144,
|
|
"learning_rate": 6.184615384615385e-07,
|
|
"loss": 0.010083096474409104,
|
|
"num_tokens": 4471416.0,
|
|
"reward": 0.7822016298770904,
|
|
"reward_std": 0.4628724604845047,
|
|
"rewards/reward_execution/mean": 0.821875,
|
|
"rewards/reward_execution/std": 0.3253744751214981,
|
|
"rewards/reward_format/mean": -0.038750000623986126,
|
|
"rewards/reward_format/std": 0.24609021246433258,
|
|
"rewards/reward_length/mean": -0.0009234109424141934,
|
|
"rewards/reward_length/std": 0.0028329493026831186,
|
|
"step": 450,
|
|
"step_time": 30.610301843500746
|
|
},
|
|
{
|
|
"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": 0.071875,
|
|
"completions/max_length": 57.9,
|
|
"completions/max_terminated_length": 55.5,
|
|
"completions/mean_length": 26.56875,
|
|
"completions/mean_terminated_length": 23.618104934692383,
|
|
"completions/min_length": 4.2,
|
|
"completions/min_terminated_length": 4.2,
|
|
"entropy": 0.01626600658037205,
|
|
"epoch": 1.2013071895424836,
|
|
"frac_reward_zero_std": 0.91875,
|
|
"grad_norm": 0.13261102139949799,
|
|
"learning_rate": 5.876923076923077e-07,
|
|
"loss": -0.0023310787975788115,
|
|
"num_tokens": 4566926.0,
|
|
"reward": 0.8538915812969208,
|
|
"reward_std": 0.3753566607832909,
|
|
"rewards/reward_execution/mean": 0.8421875,
|
|
"rewards/reward_execution/std": 0.26579051464796066,
|
|
"rewards/reward_format/mean": 0.011874999850988388,
|
|
"rewards/reward_format/std": 0.16385597586631775,
|
|
"rewards/reward_length/mean": -0.00017096186384151223,
|
|
"rewards/reward_length/std": 0.000871174709754996,
|
|
"step": 460,
|
|
"step_time": 24.040774445600256
|
|
},
|
|
{
|
|
"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": 0.13125,
|
|
"completions/max_length": 63.3,
|
|
"completions/max_terminated_length": 57.7,
|
|
"completions/mean_length": 32.83125,
|
|
"completions/mean_terminated_length": 27.75711135864258,
|
|
"completions/min_length": 7.0,
|
|
"completions/min_terminated_length": 7.0,
|
|
"entropy": 0.02020037005277118,
|
|
"epoch": 1.227450980392157,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.1261768341064453,
|
|
"learning_rate": 5.569230769230769e-07,
|
|
"loss": 0.006209623441100121,
|
|
"num_tokens": 4672236.0,
|
|
"reward": 0.8206767499446869,
|
|
"reward_std": 0.4351291760802269,
|
|
"rewards/reward_execution/mean": 0.78125,
|
|
"rewards/reward_execution/std": 0.35555928349494936,
|
|
"rewards/reward_format/mean": 0.0400000000372529,
|
|
"rewards/reward_format/std": 0.1416501671075821,
|
|
"rewards/reward_length/mean": -0.0005732660632929764,
|
|
"rewards/reward_length/std": 0.0024486880844051483,
|
|
"step": 470,
|
|
"step_time": 25.54749037789861
|
|
},
|
|
{
|
|
"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": 0.15625,
|
|
"completions/max_length": 61.3,
|
|
"completions/max_terminated_length": 52.0,
|
|
"completions/mean_length": 31.90625,
|
|
"completions/mean_terminated_length": 26.10811996459961,
|
|
"completions/min_length": 6.4,
|
|
"completions/min_terminated_length": 6.4,
|
|
"entropy": 0.021489445542647444,
|
|
"epoch": 1.25359477124183,
|
|
"frac_reward_zero_std": 0.95625,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 5.261538461538462e-07,
|
|
"loss": 0.001217818260192871,
|
|
"num_tokens": 4773142.0,
|
|
"reward": 0.7684209883213043,
|
|
"reward_std": 0.5122480720281601,
|
|
"rewards/reward_execution/mean": 0.78125,
|
|
"rewards/reward_execution/std": 0.38532922863960267,
|
|
"rewards/reward_format/mean": -0.012499999953433871,
|
|
"rewards/reward_format/std": 0.2104302704334259,
|
|
"rewards/reward_length/mean": -0.0003290332489996217,
|
|
"rewards/reward_length/std": 0.0013797201867419063,
|
|
"step": 480,
|
|
"step_time": 31.867946282702906
|
|
},
|
|
{
|
|
"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": 0.084375,
|
|
"completions/max_length": 57.5,
|
|
"completions/max_terminated_length": 54.5,
|
|
"completions/mean_length": 29.8125,
|
|
"completions/mean_terminated_length": 26.877732467651366,
|
|
"completions/min_length": 5.6,
|
|
"completions/min_terminated_length": 5.6,
|
|
"entropy": 0.018350027320229854,
|
|
"epoch": 1.2797385620915032,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.3156188726425171,
|
|
"learning_rate": 4.953846153846154e-07,
|
|
"loss": -0.002997533604502678,
|
|
"num_tokens": 4871182.0,
|
|
"reward": 0.7879133224487305,
|
|
"reward_std": 0.4440931737422943,
|
|
"rewards/reward_execution/mean": 0.759375,
|
|
"rewards/reward_execution/std": 0.38551526963710786,
|
|
"rewards/reward_format/mean": 0.028750000521540642,
|
|
"rewards/reward_format/std": 0.1575431391596794,
|
|
"rewards/reward_length/mean": -0.0002117048636137042,
|
|
"rewards/reward_length/std": 0.000824647574336268,
|
|
"step": 490,
|
|
"step_time": 24.48725758479886
|
|
},
|
|
{
|
|
"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": 0.08125,
|
|
"completions/max_length": 60.3,
|
|
"completions/max_terminated_length": 54.1,
|
|
"completions/mean_length": 24.778125,
|
|
"completions/mean_terminated_length": 21.149474239349367,
|
|
"completions/min_length": 1.7,
|
|
"completions/min_terminated_length": 1.7,
|
|
"entropy": 0.017599240622803335,
|
|
"epoch": 1.3058823529411765,
|
|
"frac_reward_zero_std": 0.91875,
|
|
"grad_norm": 0.26232999563217163,
|
|
"learning_rate": 4.646153846153846e-07,
|
|
"loss": 0.0008274193853139878,
|
|
"num_tokens": 4969239.0,
|
|
"reward": 0.7705635547637939,
|
|
"reward_std": 0.4432795435190201,
|
|
"rewards/reward_execution/mean": 0.8375,
|
|
"rewards/reward_execution/std": 0.288656148314476,
|
|
"rewards/reward_format/mean": -0.06687500057741999,
|
|
"rewards/reward_format/std": 0.24123640060424806,
|
|
"rewards/reward_length/mean": -6.14795833826065e-05,
|
|
"rewards/reward_length/std": 0.000347781041637063,
|
|
"step": 500,
|
|
"step_time": 30.19148360179897
|
|
},
|
|
{
|
|
"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": 0.03125,
|
|
"completions/max_length": 51.2,
|
|
"completions/max_terminated_length": 50.1,
|
|
"completions/mean_length": 26.78125,
|
|
"completions/mean_terminated_length": 25.568227195739745,
|
|
"completions/min_length": 5.1,
|
|
"completions/min_terminated_length": 5.1,
|
|
"entropy": 0.024345160353732354,
|
|
"epoch": 1.3320261437908496,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.5119605660438538,
|
|
"learning_rate": 4.338461538461538e-07,
|
|
"loss": 0.004315494000911713,
|
|
"num_tokens": 5062105.0,
|
|
"reward": 0.850159564614296,
|
|
"reward_std": 0.36619447022676466,
|
|
"rewards/reward_execution/mean": 0.85625,
|
|
"rewards/reward_execution/std": 0.26700153946876526,
|
|
"rewards/reward_format/mean": -0.004999999515712261,
|
|
"rewards/reward_format/std": 0.18415991365909576,
|
|
"rewards/reward_length/mean": -0.0010904625472903718,
|
|
"rewards/reward_length/std": 0.0033928818433196285,
|
|
"step": 510,
|
|
"step_time": 22.818664845402964
|
|
},
|
|
{
|
|
"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": 0.078125,
|
|
"completions/max_length": 60.9,
|
|
"completions/max_terminated_length": 57.7,
|
|
"completions/mean_length": 33.153125,
|
|
"completions/mean_terminated_length": 31.169764137268068,
|
|
"completions/min_length": 8.4,
|
|
"completions/min_terminated_length": 8.4,
|
|
"entropy": 0.02092429403019196,
|
|
"epoch": 1.358169934640523,
|
|
"frac_reward_zero_std": 0.84375,
|
|
"grad_norm": 0.3822065591812134,
|
|
"learning_rate": 4.03076923076923e-07,
|
|
"loss": -0.010526281595230103,
|
|
"num_tokens": 5174990.0,
|
|
"reward": 0.8278884023427964,
|
|
"reward_std": 0.3834263629585621,
|
|
"rewards/reward_execution/mean": 0.7828125,
|
|
"rewards/reward_execution/std": 0.32498968541622164,
|
|
"rewards/reward_format/mean": 0.04562500044703484,
|
|
"rewards/reward_format/std": 0.1361053392291069,
|
|
"rewards/reward_length/mean": -0.0005491247362442664,
|
|
"rewards/reward_length/std": 0.001976083514091442,
|
|
"step": 520,
|
|
"step_time": 31.536296672303433
|
|
},
|
|
{
|
|
"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": 0.04375,
|
|
"completions/max_length": 59.0,
|
|
"completions/max_terminated_length": 55.2,
|
|
"completions/mean_length": 28.121875,
|
|
"completions/mean_terminated_length": 26.3939302444458,
|
|
"completions/min_length": 5.8,
|
|
"completions/min_terminated_length": 5.8,
|
|
"entropy": 0.019862789660692215,
|
|
"epoch": 1.384313725490196,
|
|
"frac_reward_zero_std": 0.9625,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 3.7230769230769226e-07,
|
|
"loss": 0.003444021940231323,
|
|
"num_tokens": 5268597.0,
|
|
"reward": 0.9102885007858277,
|
|
"reward_std": 0.3575027883052826,
|
|
"rewards/reward_execution/mean": 0.8703125,
|
|
"rewards/reward_execution/std": 0.2867819845676422,
|
|
"rewards/reward_format/mean": 0.04000000082887709,
|
|
"rewards/reward_format/std": 0.11716837733983994,
|
|
"rewards/reward_length/mean": -2.4026141909416764e-05,
|
|
"rewards/reward_length/std": 9.454178507439792e-05,
|
|
"step": 530,
|
|
"step_time": 30.988930102699307
|
|
},
|
|
{
|
|
"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": 0.096875,
|
|
"completions/max_length": 60.5,
|
|
"completions/max_terminated_length": 55.4,
|
|
"completions/mean_length": 31.140625,
|
|
"completions/mean_terminated_length": 27.804740142822265,
|
|
"completions/min_length": 6.2,
|
|
"completions/min_terminated_length": 6.2,
|
|
"entropy": 0.014118772479559994,
|
|
"epoch": 1.4104575163398692,
|
|
"frac_reward_zero_std": 0.91875,
|
|
"grad_norm": 0.1555328518152237,
|
|
"learning_rate": 3.415384615384615e-07,
|
|
"loss": 0.0035601556301116943,
|
|
"num_tokens": 5363498.0,
|
|
"reward": 0.8472207427024842,
|
|
"reward_std": 0.41531313955783844,
|
|
"rewards/reward_execution/mean": 0.81875,
|
|
"rewards/reward_execution/std": 0.32633625268936156,
|
|
"rewards/reward_format/mean": 0.02874999982304871,
|
|
"rewards/reward_format/std": 0.16479601711034775,
|
|
"rewards/reward_length/mean": -0.0002792961342493072,
|
|
"rewards/reward_length/std": 0.0015470008744159713,
|
|
"step": 540,
|
|
"step_time": 23.504739268400225
|
|
},
|
|
{
|
|
"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": 0.05625,
|
|
"completions/max_length": 60.5,
|
|
"completions/max_terminated_length": 57.4,
|
|
"completions/mean_length": 30.140625,
|
|
"completions/mean_terminated_length": 27.946498298645018,
|
|
"completions/min_length": 5.7,
|
|
"completions/min_terminated_length": 5.7,
|
|
"entropy": 0.018889845718240393,
|
|
"epoch": 1.4366013071895425,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.14116764068603516,
|
|
"learning_rate": 3.1076923076923073e-07,
|
|
"loss": -0.0020487412810325624,
|
|
"num_tokens": 5460627.0,
|
|
"reward": 0.8742155015468598,
|
|
"reward_std": 0.33968957643955944,
|
|
"rewards/reward_execution/mean": 0.8421875,
|
|
"rewards/reward_execution/std": 0.25875071734189986,
|
|
"rewards/reward_format/mean": 0.032500000670552254,
|
|
"rewards/reward_format/std": 0.15005087852478027,
|
|
"rewards/reward_length/mean": -0.00047204937000060453,
|
|
"rewards/reward_length/std": 0.0026132273662369697,
|
|
"step": 550,
|
|
"step_time": 24.660869941700366
|
|
},
|
|
{
|
|
"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": 0.05,
|
|
"completions/max_length": 61.9,
|
|
"completions/max_terminated_length": 58.3,
|
|
"completions/mean_length": 28.68125,
|
|
"completions/mean_terminated_length": 26.72380485534668,
|
|
"completions/min_length": 6.7,
|
|
"completions/min_terminated_length": 6.7,
|
|
"entropy": 0.017564299988407583,
|
|
"epoch": 1.4627450980392158,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.12466259300708771,
|
|
"learning_rate": 2.8e-07,
|
|
"loss": -9.495913982391358e-05,
|
|
"num_tokens": 5553421.0,
|
|
"reward": 0.8958252310752869,
|
|
"reward_std": 0.42488523572683334,
|
|
"rewards/reward_execution/mean": 0.8375,
|
|
"rewards/reward_execution/std": 0.34917561113834383,
|
|
"rewards/reward_format/mean": 0.05875000078231096,
|
|
"rewards/reward_format/std": 0.10345140993595123,
|
|
"rewards/reward_length/mean": -0.00042480286210775373,
|
|
"rewards/reward_length/std": 0.0011976861394941806,
|
|
"step": 560,
|
|
"step_time": 21.848428737500218
|
|
},
|
|
{
|
|
"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": 0.084375,
|
|
"completions/max_length": 57.0,
|
|
"completions/max_terminated_length": 50.5,
|
|
"completions/mean_length": 27.453125,
|
|
"completions/mean_terminated_length": 24.09313259124756,
|
|
"completions/min_length": 6.5,
|
|
"completions/min_terminated_length": 6.5,
|
|
"entropy": 0.01437177338884794,
|
|
"epoch": 1.488888888888889,
|
|
"frac_reward_zero_std": 0.9125,
|
|
"grad_norm": 0.16263368725776672,
|
|
"learning_rate": 2.492307692307692e-07,
|
|
"loss": -6.0356780886650084e-05,
|
|
"num_tokens": 5641074.0,
|
|
"reward": 0.847717821598053,
|
|
"reward_std": 0.39118997007608414,
|
|
"rewards/reward_execution/mean": 0.815625,
|
|
"rewards/reward_execution/std": 0.3153843358159065,
|
|
"rewards/reward_format/mean": 0.032500000298023225,
|
|
"rewards/reward_format/std": 0.12670995891094208,
|
|
"rewards/reward_length/mean": -0.00040721593941270837,
|
|
"rewards/reward_length/std": 0.0012380244530504569,
|
|
"step": 570,
|
|
"step_time": 20.655492449797748
|
|
},
|
|
{
|
|
"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": 0.125,
|
|
"completions/max_length": 62.4,
|
|
"completions/max_terminated_length": 54.1,
|
|
"completions/mean_length": 29.6875,
|
|
"completions/mean_terminated_length": 24.902353477478027,
|
|
"completions/min_length": 5.7,
|
|
"completions/min_terminated_length": 5.7,
|
|
"entropy": 0.02779115122539224,
|
|
"epoch": 1.515032679738562,
|
|
"frac_reward_zero_std": 0.875,
|
|
"grad_norm": 0.48837918043136597,
|
|
"learning_rate": 2.1846153846153844e-07,
|
|
"loss": 0.005804136767983437,
|
|
"num_tokens": 5752042.0,
|
|
"reward": 0.7038628101348877,
|
|
"reward_std": 0.5261946886777877,
|
|
"rewards/reward_execution/mean": 0.7234375,
|
|
"rewards/reward_execution/std": 0.4040894240140915,
|
|
"rewards/reward_format/mean": -0.01812500022351742,
|
|
"rewards/reward_format/std": 0.2189276024699211,
|
|
"rewards/reward_length/mean": -0.001449720887467265,
|
|
"rewards/reward_length/std": 0.004976476434239885,
|
|
"step": 580,
|
|
"step_time": 37.17970179000404
|
|
},
|
|
{
|
|
"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": 0.05,
|
|
"completions/max_length": 56.7,
|
|
"completions/max_terminated_length": 55.0,
|
|
"completions/mean_length": 26.9875,
|
|
"completions/mean_terminated_length": 24.802696228027344,
|
|
"completions/min_length": 3.1,
|
|
"completions/min_terminated_length": 3.1,
|
|
"entropy": 0.014986539808614908,
|
|
"epoch": 1.5411764705882351,
|
|
"frac_reward_zero_std": 0.8875,
|
|
"grad_norm": 0.35831770300865173,
|
|
"learning_rate": 1.8769230769230767e-07,
|
|
"loss": -0.016387008130550385,
|
|
"num_tokens": 5864430.0,
|
|
"reward": 0.8480649173259736,
|
|
"reward_std": 0.3743706375360489,
|
|
"rewards/reward_execution/mean": 0.8484375,
|
|
"rewards/reward_execution/std": 0.28665119111537934,
|
|
"rewards/reward_format/mean": 0.000624999962747097,
|
|
"rewards/reward_format/std": 0.19347605854272842,
|
|
"rewards/reward_length/mean": -0.0009976251771149692,
|
|
"rewards/reward_length/std": 0.0034358914534095676,
|
|
"step": 590,
|
|
"step_time": 37.406214285001624
|
|
},
|
|
{
|
|
"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": 0.065625,
|
|
"completions/max_length": 61.7,
|
|
"completions/max_terminated_length": 53.8,
|
|
"completions/mean_length": 26.115625,
|
|
"completions/mean_terminated_length": 23.65874614715576,
|
|
"completions/min_length": 5.2,
|
|
"completions/min_terminated_length": 5.2,
|
|
"entropy": 0.019943572257034248,
|
|
"epoch": 1.5673202614379085,
|
|
"frac_reward_zero_std": 0.8875,
|
|
"grad_norm": 0.1319354921579361,
|
|
"learning_rate": 1.569230769230769e-07,
|
|
"loss": 0.007358292490243912,
|
|
"num_tokens": 5974659.0,
|
|
"reward": 0.8393164038658142,
|
|
"reward_std": 0.4065816730260849,
|
|
"rewards/reward_execution/mean": 0.8328125,
|
|
"rewards/reward_execution/std": 0.33312133848667147,
|
|
"rewards/reward_format/mean": 0.00812500026077032,
|
|
"rewards/reward_format/std": 0.1786261409521103,
|
|
"rewards/reward_length/mean": -0.0016211178051889874,
|
|
"rewards/reward_length/std": 0.004564436491273227,
|
|
"step": 600,
|
|
"step_time": 34.65252304400237
|
|
},
|
|
{
|
|
"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": 0.1,
|
|
"completions/max_length": 57.7,
|
|
"completions/max_terminated_length": 54.0,
|
|
"completions/mean_length": 29.984375,
|
|
"completions/mean_terminated_length": 26.44609432220459,
|
|
"completions/min_length": 6.0,
|
|
"completions/min_terminated_length": 6.0,
|
|
"entropy": 0.016775242842004445,
|
|
"epoch": 1.5934640522875818,
|
|
"frac_reward_zero_std": 0.95625,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 1.2615384615384615e-07,
|
|
"loss": 0.002294604852795601,
|
|
"num_tokens": 6098906.0,
|
|
"reward": 0.8341630756855011,
|
|
"reward_std": 0.45493687093257906,
|
|
"rewards/reward_execution/mean": 0.809375,
|
|
"rewards/reward_execution/std": 0.33004741966724394,
|
|
"rewards/reward_format/mean": 0.02500000037252903,
|
|
"rewards/reward_format/std": 0.1876816302537918,
|
|
"rewards/reward_length/mean": -0.00021195924000494415,
|
|
"rewards/reward_length/std": 0.001105904230644228,
|
|
"step": 610,
|
|
"step_time": 55.9543958102975
|
|
},
|
|
{
|
|
"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": 0.115625,
|
|
"completions/max_length": 57.1,
|
|
"completions/max_terminated_length": 53.2,
|
|
"completions/mean_length": 31.71875,
|
|
"completions/mean_terminated_length": 27.6397497177124,
|
|
"completions/min_length": 5.6,
|
|
"completions/min_terminated_length": 5.6,
|
|
"entropy": 0.01866877809916332,
|
|
"epoch": 1.619607843137255,
|
|
"frac_reward_zero_std": 0.90625,
|
|
"grad_norm": 0.09038796275854111,
|
|
"learning_rate": 9.538461538461538e-08,
|
|
"loss": -0.00958675444126129,
|
|
"num_tokens": 6198604.0,
|
|
"reward": 0.79468674659729,
|
|
"reward_std": 0.44835321605205536,
|
|
"rewards/reward_execution/mean": 0.8015625,
|
|
"rewards/reward_execution/std": 0.36473555862903595,
|
|
"rewards/reward_format/mean": -0.006874999031424522,
|
|
"rewards/reward_format/std": 0.17806818187236786,
|
|
"rewards/reward_length/mean": -7.793001714162528e-07,
|
|
"rewards/reward_length/std": 4.408387394505553e-06,
|
|
"step": 620,
|
|
"step_time": 30.469027967802322
|
|
},
|
|
{
|
|
"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": 0.040625,
|
|
"completions/max_length": 55.0,
|
|
"completions/max_terminated_length": 50.4,
|
|
"completions/mean_length": 29.18125,
|
|
"completions/mean_terminated_length": 27.752304363250733,
|
|
"completions/min_length": 7.5,
|
|
"completions/min_terminated_length": 7.5,
|
|
"entropy": 0.01569155839488303,
|
|
"epoch": 1.645751633986928,
|
|
"frac_reward_zero_std": 0.90625,
|
|
"grad_norm": 0.22198496758937836,
|
|
"learning_rate": 6.461538461538462e-08,
|
|
"loss": -0.007624600827693939,
|
|
"num_tokens": 6294410.0,
|
|
"reward": 0.9130154013633728,
|
|
"reward_std": 0.34247957170009613,
|
|
"rewards/reward_execution/mean": 0.884375,
|
|
"rewards/reward_execution/std": 0.2596578121185303,
|
|
"rewards/reward_format/mean": 0.02875000024214387,
|
|
"rewards/reward_format/std": 0.14913129806518555,
|
|
"rewards/reward_length/mean": -0.00010963492968585342,
|
|
"rewards/reward_length/std": 0.0005971741826215293,
|
|
"step": 630,
|
|
"step_time": 27.796476001703194
|
|
},
|
|
{
|
|
"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": 0.09375,
|
|
"completions/max_length": 62.2,
|
|
"completions/max_terminated_length": 57.8,
|
|
"completions/mean_length": 29.284375,
|
|
"completions/mean_terminated_length": 25.66937713623047,
|
|
"completions/min_length": 4.5,
|
|
"completions/min_terminated_length": 4.5,
|
|
"entropy": 0.01541232777399273,
|
|
"epoch": 1.6718954248366014,
|
|
"frac_reward_zero_std": 0.93125,
|
|
"grad_norm": 0.0,
|
|
"learning_rate": 3.384615384615384e-08,
|
|
"loss": -0.0036260098218917846,
|
|
"num_tokens": 6400701.0,
|
|
"reward": 0.8131250202655792,
|
|
"reward_std": 0.435266837477684,
|
|
"rewards/reward_execution/mean": 0.8125,
|
|
"rewards/reward_execution/std": 0.3499409854412079,
|
|
"rewards/reward_format/mean": 0.0006250005215406417,
|
|
"rewards/reward_format/std": 0.17473636120557784,
|
|
"rewards/reward_length/mean": 0.0,
|
|
"rewards/reward_length/std": 0.0,
|
|
"step": 640,
|
|
"step_time": 39.37825159670174
|
|
},
|
|
{
|
|
"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": 0.075,
|
|
"completions/max_length": 58.2,
|
|
"completions/max_terminated_length": 52.9,
|
|
"completions/mean_length": 27.915625,
|
|
"completions/mean_terminated_length": 24.662797927856445,
|
|
"completions/min_length": 5.5,
|
|
"completions/min_terminated_length": 5.5,
|
|
"entropy": 0.012843446540136938,
|
|
"epoch": 1.6980392156862745,
|
|
"frac_reward_zero_std": 0.9375,
|
|
"grad_norm": 0.10003364086151123,
|
|
"learning_rate": 3.076923076923077e-09,
|
|
"loss": -0.002281202748417854,
|
|
"num_tokens": 6499498.0,
|
|
"reward": 0.8679595649242401,
|
|
"reward_std": 0.3724601298570633,
|
|
"rewards/reward_execution/mean": 0.8546875,
|
|
"rewards/reward_execution/std": 0.2777053236961365,
|
|
"rewards/reward_format/mean": 0.013750000391155481,
|
|
"rewards/reward_format/std": 0.143649360537529,
|
|
"rewards/reward_length/mean": -0.00047798028826946394,
|
|
"rewards/reward_length/std": 0.0015842993328988087,
|
|
"step": 650,
|
|
"step_time": 38.02898872320402
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 650,
|
|
"num_input_tokens_seen": 6499498,
|
|
"num_train_epochs": 2,
|
|
"save_steps": 100,
|
|
"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": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|