Files
qwen2.5-7B-rlvr_g8_b384_math/trainer_state.json
ModelHub XC ab147cffe1 初始化项目,由ModelHub XC社区提供模型
Model: gguk2on/qwen2.5-7B-rlvr_g8_b384_math
Source: Original Platform
2026-04-21 11:56:04 +08:00

2264 lines
82 KiB
JSON

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.4992,
"eval_steps": 15,
"global_step": 78,
"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": 0.01953125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3781.0,
"completions/mean_length": 555.3671875,
"completions/mean_terminated_length": 484.836669921875,
"completions/min_length": 3.0,
"completions/min_terminated_length": 3.0,
"epoch": 0.0064,
"grad_norm": 0.003704642876982689,
"learning_rate": 3.125e-07,
"loss": 0.0241,
"num_tokens": 1107052.0,
"reward": 0.3880208432674408,
"reward_std": 0.3527620732784271,
"rewards/accuracy_reward": 0.236328125,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.5397135615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 1
},
{
"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": 0.01888020833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3997.0,
"completions/mean_length": 593.521484375,
"completions/mean_terminated_length": 526.1214599609375,
"completions/min_length": 11.0,
"completions/min_terminated_length": 11.0,
"epoch": 0.0128,
"grad_norm": 0.002923233201727271,
"learning_rate": 6.25e-07,
"loss": 0.0154,
"num_tokens": 2274701.0,
"reward": 0.3753255307674408,
"reward_std": 0.3523617386817932,
"rewards/accuracy_reward": 0.2428385466337204,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.5078125,
"rewards/mean_confidence_reward": 0.0,
"step": 2
},
{
"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": 0.02278645833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4065.0,
"completions/mean_length": 621.548828125,
"completions/mean_terminated_length": 540.5323486328125,
"completions/min_length": 14.0,
"completions/min_terminated_length": 14.0,
"epoch": 0.0192,
"grad_norm": 0.002838415326550603,
"learning_rate": 9.375000000000001e-07,
"loss": 0.0186,
"num_tokens": 3489440.0,
"reward": 0.4033203125,
"reward_std": 0.3530685305595398,
"rewards/accuracy_reward": 0.2526041567325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.5540364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 3
},
{
"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": 0.02213541666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4094.0,
"completions/mean_length": 532.0306396484375,
"completions/mean_terminated_length": 451.3548278808594,
"completions/min_length": 2.0,
"completions/min_terminated_length": 2.0,
"epoch": 0.0256,
"grad_norm": 0.003787894267588854,
"learning_rate": 1.25e-06,
"loss": 0.0197,
"num_tokens": 4548567.0,
"reward": 0.4485677182674408,
"reward_std": 0.35836929082870483,
"rewards/accuracy_reward": 0.2838541567325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.61328125,
"rewards/mean_confidence_reward": 0.0,
"step": 4
},
{
"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": 0.02278645833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4055.0,
"completions/mean_length": 584.3060302734375,
"completions/mean_terminated_length": 502.4210510253906,
"completions/min_length": 4.0,
"completions/min_terminated_length": 4.0,
"epoch": 0.032,
"grad_norm": 0.005179207772016525,
"learning_rate": 1.5625e-06,
"loss": 0.0199,
"num_tokens": 5699869.0,
"reward": 0.4384765625,
"reward_std": 0.34911227226257324,
"rewards/accuracy_reward": 0.2630208432674408,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.6139323115348816,
"rewards/mean_confidence_reward": 0.0,
"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": 0.02669270833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3892.0,
"completions/mean_length": 575.130859375,
"completions/mean_terminated_length": 478.5718994140625,
"completions/min_length": 7.0,
"completions/min_terminated_length": 7.0,
"epoch": 0.0384,
"grad_norm": 0.011147035285830498,
"learning_rate": 1.8750000000000003e-06,
"loss": 0.0203,
"num_tokens": 6838910.0,
"reward": 0.4508463740348816,
"reward_std": 0.3353247046470642,
"rewards/accuracy_reward": 0.25390625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.6477864384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 6
},
{
"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": 0.01822916666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4027.0,
"completions/mean_length": 502.00390625,
"completions/mean_terminated_length": 435.2718811035156,
"completions/min_length": 9.0,
"completions/min_terminated_length": 9.0,
"epoch": 0.0448,
"grad_norm": 0.01325953472405672,
"learning_rate": 2.1875000000000002e-06,
"loss": 0.023,
"num_tokens": 7867508.0,
"reward": 0.564453125,
"reward_std": 0.2995663285255432,
"rewards/accuracy_reward": 0.328125,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.80078125,
"rewards/mean_confidence_reward": 0.0,
"step": 7
},
{
"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": 0.01822916666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4003.0,
"completions/mean_length": 463.3372497558594,
"completions/mean_terminated_length": 395.88726806640625,
"completions/min_length": 12.0,
"completions/min_terminated_length": 12.0,
"epoch": 0.0512,
"grad_norm": 0.0010683703003451228,
"learning_rate": 2.5e-06,
"loss": 0.0198,
"num_tokens": 8829290.0,
"reward": 0.6012369990348816,
"reward_std": 0.26198610663414,
"rewards/accuracy_reward": 0.3209635317325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.8815104365348816,
"rewards/mean_confidence_reward": 0.0,
"step": 8
},
{
"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": 0.01106770833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3628.0,
"completions/mean_length": 421.8196716308594,
"completions/mean_terminated_length": 380.6997985839844,
"completions/min_length": 31.0,
"completions/min_terminated_length": 31.0,
"epoch": 0.0576,
"grad_norm": 0.0007295631221495569,
"learning_rate": 2.8125e-06,
"loss": 0.0142,
"num_tokens": 9735477.0,
"reward": 0.6845703125,
"reward_std": 0.24511408805847168,
"rewards/accuracy_reward": 0.4322916567325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9368489384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 9
},
{
"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": 0.01106770833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3913.0,
"completions/mean_length": 404.9270935058594,
"completions/mean_terminated_length": 363.6181640625,
"completions/min_length": 24.0,
"completions/min_terminated_length": 24.0,
"epoch": 0.064,
"grad_norm": 0.0006488566286861897,
"learning_rate": 3.125e-06,
"loss": 0.003,
"num_tokens": 10619397.0,
"reward": 0.6578776240348816,
"reward_std": 0.2086045742034912,
"rewards/accuracy_reward": 0.3502604067325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9654948115348816,
"rewards/mean_confidence_reward": 0.0,
"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": 0.01236979166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3736.0,
"completions/mean_length": 388.5774841308594,
"completions/mean_terminated_length": 342.1430358886719,
"completions/min_length": 16.0,
"completions/min_terminated_length": 16.0,
"epoch": 0.0704,
"grad_norm": 0.0006107186200097203,
"learning_rate": 3.4375e-06,
"loss": 0.0069,
"num_tokens": 11470260.0,
"reward": 0.6689453125,
"reward_std": 0.19426743686199188,
"rewards/accuracy_reward": 0.36328125,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.974609375,
"rewards/mean_confidence_reward": 0.0,
"step": 11
},
{
"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": 0.01041666666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3337.0,
"completions/mean_length": 397.7044372558594,
"completions/mean_terminated_length": 358.7750244140625,
"completions/min_length": 13.0,
"completions/min_terminated_length": 13.0,
"epoch": 0.0768,
"grad_norm": 0.0006395940436050296,
"learning_rate": 3.7500000000000005e-06,
"loss": 0.0029,
"num_tokens": 12329606.0,
"reward": 0.6666666865348816,
"reward_std": 0.19114232063293457,
"rewards/accuracy_reward": 0.3515625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9817708134651184,
"rewards/mean_confidence_reward": 0.0,
"step": 12
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3702.0,
"completions/mean_length": 359.02734375,
"completions/mean_terminated_length": 341.91888427734375,
"completions/min_length": 22.0,
"completions/min_terminated_length": 22.0,
"epoch": 0.0832,
"grad_norm": 0.000690784421749413,
"learning_rate": 4.0625000000000005e-06,
"loss": 0.0011,
"num_tokens": 13131600.0,
"reward": 0.7041015625,
"reward_std": 0.18511804938316345,
"rewards/accuracy_reward": 0.4186197817325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9895833134651184,
"rewards/mean_confidence_reward": 0.0,
"step": 13
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3094.0,
"completions/mean_length": 398.13671875,
"completions/mean_terminated_length": 381.2073059082031,
"completions/min_length": 14.0,
"completions/min_terminated_length": 14.0,
"epoch": 0.0896,
"grad_norm": 0.0004906603717245162,
"learning_rate": 4.3750000000000005e-06,
"loss": -0.001,
"num_tokens": 13995730.0,
"reward": 0.7259114980697632,
"reward_std": 0.18286210298538208,
"rewards/accuracy_reward": 0.4576822817325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.994140625,
"rewards/mean_confidence_reward": 0.0,
"step": 14
},
{
"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": 0.00846354166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4092.0,
"completions/mean_length": 436.1712341308594,
"completions/mean_terminated_length": 404.9317321777344,
"completions/min_length": 42.0,
"completions/min_terminated_length": 42.0,
"epoch": 0.096,
"grad_norm": 0.00048343793605454266,
"learning_rate": 4.6875000000000004e-06,
"loss": 0.0086,
"num_tokens": 14925649.0,
"reward": 0.7399088740348816,
"reward_std": 0.1765557825565338,
"rewards/accuracy_reward": 0.4928385317325592,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9869791865348816,
"rewards/mean_confidence_reward": 0.0,
"step": 15
},
{
"epoch": 0.096,
"eval_completions/clipped_ratio": 0.0068359375,
"eval_completions/max_length": 3393.75,
"eval_completions/max_terminated_length": 1931.25,
"eval_completions/mean_length": 430.47265625,
"eval_completions/mean_terminated_length": 405.194766998291,
"eval_completions/min_length": 71.0,
"eval_completions/min_terminated_length": 71.0,
"eval_loss": 0.0,
"eval_num_tokens": 14925649.0,
"eval_reward": 0.73095703125,
"eval_reward_std": 0.2545679435133934,
"eval_rewards/accuracy_reward": 0.4697265625,
"eval_rewards/brier_reward": 0.0,
"eval_rewards/confidence_one_or_zero": 0.0,
"eval_rewards/format_reward": 0.9921875,
"eval_rewards/mean_confidence_reward": 0.0,
"eval_runtime": 219.6084,
"eval_samples_per_second": 4.554,
"eval_steps_per_second": 0.036,
"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": 0.005859375,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4018.0,
"completions/mean_length": 426.626953125,
"completions/mean_terminated_length": 405.0,
"completions/min_length": 26.0,
"completions/min_terminated_length": 26.0,
"epoch": 0.1024,
"grad_norm": 0.003415175247937441,
"learning_rate": 5e-06,
"loss": 0.0059,
"num_tokens": 15840852.0,
"reward": 0.7454427480697632,
"reward_std": 0.18521976470947266,
"rewards/accuracy_reward": 0.5013020634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9895833134651184,
"rewards/mean_confidence_reward": 0.0,
"step": 16
},
{
"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": 0.013671875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4036.0,
"completions/mean_length": 487.5833435058594,
"completions/mean_terminated_length": 437.5657043457031,
"completions/min_length": 58.0,
"completions/min_terminated_length": 58.0,
"epoch": 0.1088,
"grad_norm": 0.014726120978593826,
"learning_rate": 4.919354838709678e-06,
"loss": 0.0066,
"num_tokens": 16847004.0,
"reward": 0.7158203125,
"reward_std": 0.17167815566062927,
"rewards/accuracy_reward": 0.44921875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.982421875,
"rewards/mean_confidence_reward": 0.0,
"step": 17
},
{
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3088.0,
"completions/mean_length": 502.166015625,
"completions/mean_terminated_length": 476.2432861328125,
"completions/min_length": 69.0,
"completions/min_terminated_length": 69.0,
"epoch": 0.1152,
"grad_norm": 0.0003558692696969956,
"learning_rate": 4.838709677419355e-06,
"loss": 0.0032,
"num_tokens": 17866827.0,
"reward": 0.7649739980697632,
"reward_std": 0.1702839434146881,
"rewards/accuracy_reward": 0.5390625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9908854365348816,
"rewards/mean_confidence_reward": 0.0,
"step": 18
},
{
"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": 0.01236979166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4024.0,
"completions/mean_length": 523.8353271484375,
"completions/mean_terminated_length": 479.09490966796875,
"completions/min_length": 52.0,
"completions/min_terminated_length": 52.0,
"epoch": 0.1216,
"grad_norm": 0.000376895593944937,
"learning_rate": 4.758064516129033e-06,
"loss": 0.0101,
"num_tokens": 18925582.0,
"reward": 0.755859375,
"reward_std": 0.14837318658828735,
"rewards/accuracy_reward": 0.5279948115348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9837239384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 19
},
{
"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": 0.01171875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3952.0,
"completions/mean_length": 555.5970458984375,
"completions/mean_terminated_length": 513.615966796875,
"completions/min_length": 95.0,
"completions/min_terminated_length": 95.0,
"epoch": 0.128,
"grad_norm": 0.00034090346889570355,
"learning_rate": 4.67741935483871e-06,
"loss": 0.0081,
"num_tokens": 20030195.0,
"reward": 0.7639974355697632,
"reward_std": 0.1521807610988617,
"rewards/accuracy_reward": 0.5423176884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9856770634651184,
"rewards/mean_confidence_reward": 0.0,
"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": 0.01302083333333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4010.0,
"completions/mean_length": 614.029296875,
"completions/mean_terminated_length": 568.093017578125,
"completions/min_length": 95.0,
"completions/min_terminated_length": 95.0,
"epoch": 0.1344,
"grad_norm": 0.00031957231112755835,
"learning_rate": 4.596774193548387e-06,
"loss": 0.0125,
"num_tokens": 21227192.0,
"reward": 0.7877604365348816,
"reward_std": 0.15472644567489624,
"rewards/accuracy_reward": 0.58984375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9856770634651184,
"rewards/mean_confidence_reward": 0.0,
"step": 21
},
{
"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": 0.009765625,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3994.0,
"completions/mean_length": 574.828125,
"completions/mean_terminated_length": 540.1026000976562,
"completions/min_length": 119.0,
"completions/min_terminated_length": 119.0,
"epoch": 0.1408,
"grad_norm": 0.0002684423525352031,
"learning_rate": 4.516129032258065e-06,
"loss": 0.0087,
"num_tokens": 22359008.0,
"reward": 0.7802734375,
"reward_std": 0.13622406125068665,
"rewards/accuracy_reward": 0.57421875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.986328125,
"rewards/mean_confidence_reward": 0.0,
"step": 22
},
{
"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": 0.01888020833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4081.0,
"completions/mean_length": 671.7708740234375,
"completions/mean_terminated_length": 605.8765869140625,
"completions/min_length": 105.0,
"completions/min_terminated_length": 105.0,
"epoch": 0.1472,
"grad_norm": 0.00025072344578802586,
"learning_rate": 4.435483870967742e-06,
"loss": 0.0164,
"num_tokens": 23650904.0,
"reward": 0.7786458730697632,
"reward_std": 0.16046416759490967,
"rewards/accuracy_reward": 0.5807291865348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9765625,
"rewards/mean_confidence_reward": 0.0,
"step": 23
},
{
"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": 0.00651041666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3210.0,
"completions/mean_length": 593.3073120117188,
"completions/mean_terminated_length": 570.3538818359375,
"completions/min_length": 75.0,
"completions/min_terminated_length": 75.0,
"epoch": 0.1536,
"grad_norm": 0.0002969923953060061,
"learning_rate": 4.35483870967742e-06,
"loss": 0.0055,
"num_tokens": 24813128.0,
"reward": 0.8098958730697632,
"reward_std": 0.12023092806339264,
"rewards/accuracy_reward": 0.6263020634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9934895634651184,
"rewards/mean_confidence_reward": 0.0,
"step": 24
},
{
"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": 0.02083333333333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4068.0,
"completions/mean_length": 643.1803588867188,
"completions/mean_terminated_length": 569.716064453125,
"completions/min_length": 80.0,
"completions/min_terminated_length": 80.0,
"epoch": 0.16,
"grad_norm": 0.0002521358255762607,
"learning_rate": 4.274193548387097e-06,
"loss": 0.0127,
"num_tokens": 26048253.0,
"reward": 0.7884114980697632,
"reward_std": 0.14166441559791565,
"rewards/accuracy_reward": 0.6002604365348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9765625,
"rewards/mean_confidence_reward": 0.0,
"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": 0.017578125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4047.0,
"completions/mean_length": 626.1217651367188,
"completions/mean_terminated_length": 564.0364990234375,
"completions/min_length": 121.0,
"completions/min_terminated_length": 121.0,
"epoch": 0.1664,
"grad_norm": 0.0002549105847720057,
"learning_rate": 4.193548387096774e-06,
"loss": 0.0126,
"num_tokens": 27265200.0,
"reward": 0.77734375,
"reward_std": 0.13510474562644958,
"rewards/accuracy_reward": 0.5735676884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9811198115348816,
"rewards/mean_confidence_reward": 0.0,
"step": 26
},
{
"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": 0.017578125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4089.0,
"completions/mean_length": 604.61328125,
"completions/mean_terminated_length": 542.1431884765625,
"completions/min_length": 98.0,
"completions/min_terminated_length": 98.0,
"epoch": 0.1728,
"grad_norm": 0.0002475589863024652,
"learning_rate": 4.112903225806452e-06,
"loss": 0.0144,
"num_tokens": 28439614.0,
"reward": 0.8125,
"reward_std": 0.13454118371009827,
"rewards/accuracy_reward": 0.64453125,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.98046875,
"rewards/mean_confidence_reward": 0.0,
"step": 27
},
{
"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": 0.01106770833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4020.0,
"completions/mean_length": 641.990234375,
"completions/mean_terminated_length": 603.3344116210938,
"completions/min_length": 159.0,
"completions/min_terminated_length": 159.0,
"epoch": 0.1792,
"grad_norm": 0.0002359977224841714,
"learning_rate": 4.032258064516129e-06,
"loss": 0.0105,
"num_tokens": 29677791.0,
"reward": 0.8151041865348816,
"reward_std": 0.13174818456172943,
"rewards/accuracy_reward": 0.6438801884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.986328125,
"rewards/mean_confidence_reward": 0.0,
"step": 28
},
{
"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": 0.01888020833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3084.0,
"completions/mean_length": 661.5521240234375,
"completions/mean_terminated_length": 595.461181640625,
"completions/min_length": 122.0,
"completions/min_terminated_length": 122.0,
"epoch": 0.1856,
"grad_norm": 0.0003096568398177624,
"learning_rate": 3.951612903225807e-06,
"loss": 0.0154,
"num_tokens": 30938983.0,
"reward": 0.7978515625,
"reward_std": 0.12398765981197357,
"rewards/accuracy_reward": 0.615234375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.98046875,
"rewards/mean_confidence_reward": 0.0,
"step": 29
},
{
"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": 0.01432291666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3647.0,
"completions/mean_length": 658.087890625,
"completions/mean_terminated_length": 608.1314697265625,
"completions/min_length": 47.0,
"completions/min_terminated_length": 47.0,
"epoch": 0.192,
"grad_norm": 0.0002497898240108043,
"learning_rate": 3.870967741935484e-06,
"loss": 0.0098,
"num_tokens": 32198790.0,
"reward": 0.787109375,
"reward_std": 0.12306138128042221,
"rewards/accuracy_reward": 0.5904948115348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9837239384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 30
},
{
"epoch": 0.192,
"eval_completions/clipped_ratio": 0.015850360576923073,
"eval_completions/max_length": 4096.0,
"eval_completions/max_terminated_length": 2561.25,
"eval_completions/mean_length": 643.6061477661133,
"eval_completions/mean_terminated_length": 587.8973083496094,
"eval_completions/min_length": 166.5,
"eval_completions/min_terminated_length": 166.5,
"eval_loss": 0.0,
"eval_num_tokens": 32198790.0,
"eval_reward": 0.80517578125,
"eval_reward_std": 0.25961557030677795,
"eval_rewards/accuracy_reward": 0.6298828125,
"eval_rewards/brier_reward": 0.0,
"eval_rewards/confidence_one_or_zero": 0.0,
"eval_rewards/format_reward": 0.98046875,
"eval_rewards/mean_confidence_reward": 0.0,
"eval_runtime": 272.2745,
"eval_samples_per_second": 3.673,
"eval_steps_per_second": 0.029,
"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": 0.01302083333333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3641.0,
"completions/mean_length": 638.5384521484375,
"completions/mean_terminated_length": 592.9254760742188,
"completions/min_length": 161.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.1984,
"grad_norm": 0.0002642054751049727,
"learning_rate": 3.7903225806451614e-06,
"loss": 0.0083,
"num_tokens": 33435001.0,
"reward": 0.8219401240348816,
"reward_std": 0.12539035081863403,
"rewards/accuracy_reward": 0.66015625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9837239384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 31
},
{
"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": 0.01432291666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3878.0,
"completions/mean_length": 673.169921875,
"completions/mean_terminated_length": 623.4326782226562,
"completions/min_length": 135.0,
"completions/min_terminated_length": 135.0,
"epoch": 0.2048,
"grad_norm": 0.00023612201039213687,
"learning_rate": 3.7096774193548392e-06,
"loss": 0.0122,
"num_tokens": 34716198.0,
"reward": 0.8108724355697632,
"reward_std": 0.135454922914505,
"rewards/accuracy_reward": 0.6373698115348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.984375,
"rewards/mean_confidence_reward": 0.0,
"step": 32
},
{
"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": 0.01236979166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2475.0,
"completions/mean_length": 600.5631713867188,
"completions/mean_terminated_length": 556.7837524414062,
"completions/min_length": 91.0,
"completions/min_terminated_length": 91.0,
"epoch": 0.2112,
"grad_norm": 0.0002798466884996742,
"learning_rate": 3.6290322580645166e-06,
"loss": 0.0093,
"num_tokens": 35888359.0,
"reward": 0.8180338740348816,
"reward_std": 0.1216612458229065,
"rewards/accuracy_reward": 0.6490885615348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9869791865348816,
"rewards/mean_confidence_reward": 0.0,
"step": 33
},
{
"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": 0.01171875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3657.0,
"completions/mean_length": 640.6198120117188,
"completions/mean_terminated_length": 599.6469116210938,
"completions/min_length": 94.0,
"completions/min_terminated_length": 94.0,
"epoch": 0.2176,
"grad_norm": 0.00022835738491266966,
"learning_rate": 3.548387096774194e-06,
"loss": 0.0067,
"num_tokens": 37122367.0,
"reward": 0.8043619990348816,
"reward_std": 0.11845268309116364,
"rewards/accuracy_reward": 0.6204426884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.98828125,
"rewards/mean_confidence_reward": 0.0,
"step": 34
},
{
"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": 0.01041666666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2511.0,
"completions/mean_length": 613.96484375,
"completions/mean_terminated_length": 577.3118896484375,
"completions/min_length": 90.0,
"completions/min_terminated_length": 90.0,
"epoch": 0.224,
"grad_norm": 0.00022982119116932154,
"learning_rate": 3.4677419354838714e-06,
"loss": 0.0074,
"num_tokens": 38318185.0,
"reward": 0.8128255605697632,
"reward_std": 0.11442983150482178,
"rewards/accuracy_reward": 0.63671875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9889323115348816,
"rewards/mean_confidence_reward": 0.0,
"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": 0.005859375,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2985.0,
"completions/mean_length": 633.0618896484375,
"completions/mean_terminated_length": 612.651611328125,
"completions/min_length": 102.0,
"completions/min_terminated_length": 102.0,
"epoch": 0.2304,
"grad_norm": 0.0002401992242084816,
"learning_rate": 3.3870967741935484e-06,
"loss": 0.004,
"num_tokens": 39545168.0,
"reward": 0.8138021230697632,
"reward_std": 0.11124514043331146,
"rewards/accuracy_reward": 0.634765625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 36
},
{
"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": 0.01106770833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3188.0,
"completions/mean_length": 644.42578125,
"completions/mean_terminated_length": 605.7972412109375,
"completions/min_length": 134.0,
"completions/min_terminated_length": 134.0,
"epoch": 0.2368,
"grad_norm": 0.0002481676929164678,
"learning_rate": 3.306451612903226e-06,
"loss": 0.008,
"num_tokens": 40785846.0,
"reward": 0.8229166865348816,
"reward_std": 0.11275224387645721,
"rewards/accuracy_reward": 0.6575520634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.98828125,
"rewards/mean_confidence_reward": 0.0,
"step": 37
},
{
"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": 0.01692708333333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3881.0,
"completions/mean_length": 656.0540771484375,
"completions/mean_terminated_length": 596.8231811523438,
"completions/min_length": 144.0,
"completions/min_terminated_length": 144.0,
"epoch": 0.2432,
"grad_norm": 0.00023255293490365148,
"learning_rate": 3.225806451612903e-06,
"loss": 0.0092,
"num_tokens": 42046369.0,
"reward": 0.81640625,
"reward_std": 0.11946391314268112,
"rewards/accuracy_reward": 0.650390625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.982421875,
"rewards/mean_confidence_reward": 0.0,
"step": 38
},
{
"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": 0.01627604166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3904.0,
"completions/mean_length": 680.505859375,
"completions/mean_terminated_length": 623.995361328125,
"completions/min_length": 122.0,
"completions/min_terminated_length": 122.0,
"epoch": 0.2496,
"grad_norm": 0.0002798327477648854,
"learning_rate": 3.145161290322581e-06,
"loss": 0.0105,
"num_tokens": 43341162.0,
"reward": 0.8089193105697632,
"reward_std": 0.13103950023651123,
"rewards/accuracy_reward": 0.63671875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9811198115348816,
"rewards/mean_confidence_reward": 0.0,
"step": 39
},
{
"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": 0.00911458333333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3231.0,
"completions/mean_length": 649.806640625,
"completions/mean_terminated_length": 618.1070556640625,
"completions/min_length": 122.0,
"completions/min_terminated_length": 122.0,
"epoch": 0.256,
"grad_norm": 0.00021782911790069193,
"learning_rate": 3.0645161290322584e-06,
"loss": 0.0071,
"num_tokens": 44593809.0,
"reward": 0.8167318105697632,
"reward_std": 0.10512678325176239,
"rewards/accuracy_reward": 0.6438801884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9895833134651184,
"rewards/mean_confidence_reward": 0.0,
"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": 0.00846354166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3253.0,
"completions/mean_length": 646.7916870117188,
"completions/mean_terminated_length": 617.3499755859375,
"completions/min_length": 151.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.2624,
"grad_norm": 0.00023386659449897707,
"learning_rate": 2.983870967741936e-06,
"loss": 0.0068,
"num_tokens": 45839065.0,
"reward": 0.833984375,
"reward_std": 0.11914779245853424,
"rewards/accuracy_reward": 0.6764323115348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9915364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 41
},
{
"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": 0.017578125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4076.0,
"completions/mean_length": 654.45703125,
"completions/mean_terminated_length": 592.8787231445312,
"completions/min_length": 104.0,
"completions/min_terminated_length": 104.0,
"epoch": 0.2688,
"grad_norm": 0.0002567242190707475,
"learning_rate": 2.903225806451613e-06,
"loss": 0.0088,
"num_tokens": 47095759.0,
"reward": 0.8098958730697632,
"reward_std": 0.11656266450881958,
"rewards/accuracy_reward": 0.6380208134651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9817708134651184,
"rewards/mean_confidence_reward": 0.0,
"step": 42
},
{
"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": 0.00651041666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3825.0,
"completions/mean_length": 653.9583740234375,
"completions/mean_terminated_length": 631.4024047851562,
"completions/min_length": 144.0,
"completions/min_terminated_length": 144.0,
"epoch": 0.2752,
"grad_norm": 0.00028055167058482766,
"learning_rate": 2.822580645161291e-06,
"loss": 0.0062,
"num_tokens": 48353599.0,
"reward": 0.8131510615348816,
"reward_std": 0.12230964004993439,
"rewards/accuracy_reward": 0.6334635615348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 43
},
{
"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": 0.0078125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4042.0,
"completions/mean_length": 624.6744995117188,
"completions/mean_terminated_length": 597.3411865234375,
"completions/min_length": 150.0,
"completions/min_terminated_length": 150.0,
"epoch": 0.2816,
"grad_norm": 0.00026211561635136604,
"learning_rate": 2.7419354838709676e-06,
"loss": 0.0065,
"num_tokens": 49565811.0,
"reward": 0.8356119990348816,
"reward_std": 0.11118366569280624,
"rewards/accuracy_reward": 0.6796875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9915364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 44
},
{
"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": 0.00651041666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3697.0,
"completions/mean_length": 641.7838745117188,
"completions/mean_terminated_length": 619.1481323242188,
"completions/min_length": 182.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.288,
"grad_norm": 0.0002798437199089676,
"learning_rate": 2.6612903225806454e-06,
"loss": 0.0046,
"num_tokens": 50801927.0,
"reward": 0.810546875,
"reward_std": 0.10706674307584763,
"rewards/accuracy_reward": 0.6282551884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 45
},
{
"epoch": 0.288,
"eval_completions/clipped_ratio": 0.010967548076923073,
"eval_completions/max_length": 3628.5,
"eval_completions/max_terminated_length": 2420.75,
"eval_completions/mean_length": 656.3849182128906,
"eval_completions/mean_terminated_length": 618.1777877807617,
"eval_completions/min_length": 171.125,
"eval_completions/min_terminated_length": 171.125,
"eval_loss": 0.0,
"eval_num_tokens": 50801927.0,
"eval_reward": 0.81494140625,
"eval_reward_std": 0.2401072047650814,
"eval_rewards/accuracy_reward": 0.6416015625,
"eval_rewards/brier_reward": 0.0,
"eval_rewards/confidence_one_or_zero": 0.0,
"eval_rewards/format_reward": 0.98828125,
"eval_rewards/mean_confidence_reward": 0.0,
"eval_runtime": 239.4014,
"eval_samples_per_second": 4.177,
"eval_steps_per_second": 0.033,
"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": 0.009765625,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3457.0,
"completions/mean_length": 617.0084838867188,
"completions/mean_terminated_length": 582.6989135742188,
"completions/min_length": 115.0,
"completions/min_terminated_length": 115.0,
"epoch": 0.2944,
"grad_norm": 0.0002833885373547673,
"learning_rate": 2.580645161290323e-06,
"loss": 0.0067,
"num_tokens": 52007356.0,
"reward": 0.826171875,
"reward_std": 0.11079498380422592,
"rewards/accuracy_reward": 0.662109375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.990234375,
"rewards/mean_confidence_reward": 0.0,
"step": 46
},
{
"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": 0.013671875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3737.0,
"completions/mean_length": 685.18359375,
"completions/mean_terminated_length": 637.9049682617188,
"completions/min_length": 149.0,
"completions/min_terminated_length": 149.0,
"epoch": 0.3008,
"grad_norm": 0.00024410766491200775,
"learning_rate": 2.5e-06,
"loss": 0.0087,
"num_tokens": 53313750.0,
"reward": 0.8173828125,
"reward_std": 0.1166023537516594,
"rewards/accuracy_reward": 0.6484375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.986328125,
"rewards/mean_confidence_reward": 0.0,
"step": 47
},
{
"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": 0.01171875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4091.0,
"completions/mean_length": 689.576171875,
"completions/mean_terminated_length": 649.1837768554688,
"completions/min_length": 123.0,
"completions/min_terminated_length": 123.0,
"epoch": 0.3072,
"grad_norm": 0.00028568808920681477,
"learning_rate": 2.4193548387096776e-06,
"loss": 0.0072,
"num_tokens": 54632059.0,
"reward": 0.7998046875,
"reward_std": 0.12417584657669067,
"rewards/accuracy_reward": 0.61328125,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.986328125,
"rewards/mean_confidence_reward": 0.0,
"step": 48
},
{
"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": 0.00390625,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3823.0,
"completions/mean_length": 627.0716552734375,
"completions/mean_terminated_length": 613.468017578125,
"completions/min_length": 138.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.3136,
"grad_norm": 0.00025566882686689496,
"learning_rate": 2.338709677419355e-06,
"loss": 0.0036,
"num_tokens": 55851281.0,
"reward": 0.8310546875,
"reward_std": 0.10658612847328186,
"rewards/accuracy_reward": 0.6686198115348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9934895634651184,
"rewards/mean_confidence_reward": 0.0,
"step": 49
},
{
"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": 0.00651041666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2449.0,
"completions/mean_length": 622.958984375,
"completions/mean_terminated_length": 600.1998901367188,
"completions/min_length": 117.0,
"completions/min_terminated_length": 117.0,
"epoch": 0.32,
"grad_norm": 0.00041423627408221364,
"learning_rate": 2.2580645161290324e-06,
"loss": 0.0078,
"num_tokens": 57059514.0,
"reward": 0.8235677480697632,
"reward_std": 0.11244156956672668,
"rewards/accuracy_reward": 0.654296875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"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": 0.01171875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2564.0,
"completions/mean_length": 636.1920776367188,
"completions/mean_terminated_length": 595.1666870117188,
"completions/min_length": 139.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.3264,
"grad_norm": 0.00027631345437839627,
"learning_rate": 2.17741935483871e-06,
"loss": 0.0093,
"num_tokens": 58286153.0,
"reward": 0.8111979365348816,
"reward_std": 0.12525992095470428,
"rewards/accuracy_reward": 0.6341145634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.98828125,
"rewards/mean_confidence_reward": 0.0,
"step": 51
},
{
"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": 0.01041666666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4094.0,
"completions/mean_length": 617.8568115234375,
"completions/mean_terminated_length": 581.2447509765625,
"completions/min_length": 141.0,
"completions/min_terminated_length": 141.0,
"epoch": 0.3328,
"grad_norm": 0.00026788286049850285,
"learning_rate": 2.096774193548387e-06,
"loss": 0.0075,
"num_tokens": 59491837.0,
"reward": 0.8157552480697632,
"reward_std": 0.10741822421550751,
"rewards/accuracy_reward": 0.642578125,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9889323115348816,
"rewards/mean_confidence_reward": 0.0,
"step": 52
},
{
"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": 0.001953125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2485.0,
"completions/mean_length": 622.8698120117188,
"completions/mean_terminated_length": 616.0730590820312,
"completions/min_length": 149.0,
"completions/min_terminated_length": 149.0,
"epoch": 0.3392,
"grad_norm": 0.000268914649495855,
"learning_rate": 2.0161290322580646e-06,
"loss": 0.0051,
"num_tokens": 60697301.0,
"reward": 0.8232421875,
"reward_std": 0.10318265855312347,
"rewards/accuracy_reward": 0.6484375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.998046875,
"rewards/mean_confidence_reward": 0.0,
"step": 53
},
{
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4086.0,
"completions/mean_length": 623.26953125,
"completions/mean_terminated_length": 598.2203369140625,
"completions/min_length": 151.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.3456,
"grad_norm": 0.0002533692750148475,
"learning_rate": 1.935483870967742e-06,
"loss": 0.0066,
"num_tokens": 61905635.0,
"reward": 0.8434244990348816,
"reward_std": 0.10225149244070053,
"rewards/accuracy_reward": 0.6946614384651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9921875,
"rewards/mean_confidence_reward": 0.0,
"step": 54
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3843.0,
"completions/mean_length": 648.0703125,
"completions/mean_terminated_length": 632.28515625,
"completions/min_length": 150.0,
"completions/min_terminated_length": 150.0,
"epoch": 0.352,
"grad_norm": 0.0002154428220819682,
"learning_rate": 1.8548387096774196e-06,
"loss": 0.004,
"num_tokens": 63155519.0,
"reward": 0.8271484375,
"reward_std": 0.09687401354312897,
"rewards/accuracy_reward": 0.6595051884651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9947916865348816,
"rewards/mean_confidence_reward": 0.0,
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3433.0,
"completions/mean_length": 703.6119995117188,
"completions/mean_terminated_length": 679.1422729492188,
"completions/min_length": 198.0,
"completions/min_terminated_length": 198.0,
"epoch": 0.3584,
"grad_norm": 0.0002669694658834487,
"learning_rate": 1.774193548387097e-06,
"loss": 0.0069,
"num_tokens": 64491107.0,
"reward": 0.8330078125,
"reward_std": 0.11356484144926071,
"rewards/accuracy_reward": 0.6731770634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 56
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4061.0,
"completions/mean_length": 618.265625,
"completions/mean_terminated_length": 602.343994140625,
"completions/min_length": 108.0,
"completions/min_terminated_length": 108.0,
"epoch": 0.3648,
"grad_norm": 0.0002474285429343581,
"learning_rate": 1.6935483870967742e-06,
"loss": 0.0065,
"num_tokens": 65693683.0,
"reward": 0.8564453125,
"reward_std": 0.10922061651945114,
"rewards/accuracy_reward": 0.71875,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.994140625,
"rewards/mean_confidence_reward": 0.0,
"step": 57
},
{
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3299.0,
"completions/mean_length": 657.6087646484375,
"completions/mean_terminated_length": 632.8071899414062,
"completions/min_length": 156.0,
"completions/min_terminated_length": 156.0,
"epoch": 0.3712,
"grad_norm": 0.0002707333769649267,
"learning_rate": 1.6129032258064516e-06,
"loss": 0.0061,
"num_tokens": 66959650.0,
"reward": 0.8492838740348816,
"reward_std": 0.11089633405208588,
"rewards/accuracy_reward": 0.7057291865348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 58
},
{
"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": 0.005859375,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3742.0,
"completions/mean_length": 635.080078125,
"completions/mean_terminated_length": 614.6817626953125,
"completions/min_length": 137.0,
"completions/min_terminated_length": 137.0,
"epoch": 0.3776,
"grad_norm": 0.00023831303406041116,
"learning_rate": 1.5322580645161292e-06,
"loss": 0.0042,
"num_tokens": 68185781.0,
"reward": 0.8180338740348816,
"reward_std": 0.0983486920595169,
"rewards/accuracy_reward": 0.6419270634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.994140625,
"rewards/mean_confidence_reward": 0.0,
"step": 59
},
{
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2337.0,
"completions/mean_length": 638.1764526367188,
"completions/mean_terminated_length": 613.2347412109375,
"completions/min_length": 159.0,
"completions/min_terminated_length": 159.0,
"epoch": 0.384,
"grad_norm": 0.00025369905051775277,
"learning_rate": 1.4516129032258066e-06,
"loss": 0.0069,
"num_tokens": 69419524.0,
"reward": 0.8206380605697632,
"reward_std": 0.10652142763137817,
"rewards/accuracy_reward": 0.6490885615348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9921875,
"rewards/mean_confidence_reward": 0.0,
"step": 60
},
{
"epoch": 0.384,
"eval_completions/clipped_ratio": 0.004131610576923073,
"eval_completions/max_length": 3169.875,
"eval_completions/max_terminated_length": 2246.5,
"eval_completions/mean_length": 643.4401321411133,
"eval_completions/mean_terminated_length": 629.0936508178711,
"eval_completions/min_length": 188.5,
"eval_completions/min_terminated_length": 188.5,
"eval_loss": 0.0,
"eval_num_tokens": 69419524.0,
"eval_reward": 0.82373046875,
"eval_reward_std": 0.23457887582480907,
"eval_rewards/accuracy_reward": 0.6513671875,
"eval_rewards/brier_reward": 0.0,
"eval_rewards/confidence_one_or_zero": 0.0,
"eval_rewards/format_reward": 0.99609375,
"eval_rewards/mean_confidence_reward": 0.0,
"eval_runtime": 209.489,
"eval_samples_per_second": 4.774,
"eval_steps_per_second": 0.038,
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2865.0,
"completions/mean_length": 673.5443115234375,
"completions/mean_terminated_length": 648.8577270507812,
"completions/min_length": 155.0,
"completions/min_terminated_length": 155.0,
"epoch": 0.3904,
"grad_norm": 0.00026515661738812923,
"learning_rate": 1.3709677419354838e-06,
"loss": 0.0045,
"num_tokens": 70716608.0,
"reward": 0.8011068105697632,
"reward_std": 0.10466423630714417,
"rewards/accuracy_reward": 0.6106770634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9915364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 61
},
{
"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": 0.00846354166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4062.0,
"completions/mean_length": 686.5358276367188,
"completions/mean_terminated_length": 657.433349609375,
"completions/min_length": 138.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.3968,
"grad_norm": 0.00023534742649644613,
"learning_rate": 1.2903225806451614e-06,
"loss": 0.0055,
"num_tokens": 72025423.0,
"reward": 0.826171875,
"reward_std": 0.10892236232757568,
"rewards/accuracy_reward": 0.662109375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.990234375,
"rewards/mean_confidence_reward": 0.0,
"step": 62
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3982.0,
"completions/mean_length": 629.6354370117188,
"completions/mean_terminated_length": 613.7658081054688,
"completions/min_length": 114.0,
"completions/min_terminated_length": 114.0,
"epoch": 0.4032,
"grad_norm": 0.00023829005658626556,
"learning_rate": 1.2096774193548388e-06,
"loss": 0.001,
"num_tokens": 73244799.0,
"reward": 0.8430989980697632,
"reward_std": 0.09954534471035004,
"rewards/accuracy_reward": 0.6927083134651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9934895634651184,
"rewards/mean_confidence_reward": 0.0,
"step": 63
},
{
"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": 0.00846354166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2787.0,
"completions/mean_length": 669.8873901367188,
"completions/mean_terminated_length": 640.642822265625,
"completions/min_length": 168.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.4096,
"grad_norm": 0.00023246412456501275,
"learning_rate": 1.1290322580645162e-06,
"loss": 0.007,
"num_tokens": 74524818.0,
"reward": 0.8408203125,
"reward_std": 0.11024834215641022,
"rewards/accuracy_reward": 0.6901041865348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9915364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 64
},
{
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3901.0,
"completions/mean_length": 669.7200927734375,
"completions/mean_terminated_length": 645.0059204101562,
"completions/min_length": 131.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.416,
"grad_norm": 0.0002526069583836943,
"learning_rate": 1.0483870967741936e-06,
"loss": 0.005,
"num_tokens": 75811244.0,
"reward": 0.8414713740348816,
"reward_std": 0.10166700929403305,
"rewards/accuracy_reward": 0.6901041865348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 65
},
{
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00651041666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3167.0,
"completions/mean_length": 673.7623901367188,
"completions/mean_terminated_length": 651.336181640625,
"completions/min_length": 175.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.4224,
"grad_norm": 0.00027628280804492533,
"learning_rate": 9.67741935483871e-07,
"loss": 0.0066,
"num_tokens": 77098391.0,
"reward": 0.8313802480697632,
"reward_std": 0.11165839433670044,
"rewards/accuracy_reward": 0.6692708134651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9934895634651184,
"rewards/mean_confidence_reward": 0.0,
"step": 66
},
{
"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": 0.009765625,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3942.0,
"completions/mean_length": 667.787109375,
"completions/mean_terminated_length": 633.9783325195312,
"completions/min_length": 149.0,
"completions/min_terminated_length": 149.0,
"epoch": 0.4288,
"grad_norm": 0.00024809566093608737,
"learning_rate": 8.870967741935485e-07,
"loss": 0.0071,
"num_tokens": 78374368.0,
"reward": 0.8408203125,
"reward_std": 0.11645618081092834,
"rewards/accuracy_reward": 0.69140625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.990234375,
"rewards/mean_confidence_reward": 0.0,
"step": 67
},
{
"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": 0.00651041666666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4054.0,
"completions/mean_length": 649.001953125,
"completions/mean_terminated_length": 626.4135131835938,
"completions/min_length": 157.0,
"completions/min_terminated_length": 157.0,
"epoch": 0.4352,
"grad_norm": 0.000251438410487026,
"learning_rate": 8.064516129032258e-07,
"loss": 0.0059,
"num_tokens": 79626331.0,
"reward": 0.8284505605697632,
"reward_std": 0.10573884099721909,
"rewards/accuracy_reward": 0.6653645634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9915364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 68
},
{
"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": 0.01171875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3946.0,
"completions/mean_length": 653.7532958984375,
"completions/mean_terminated_length": 612.9360961914062,
"completions/min_length": 129.0,
"completions/min_terminated_length": 129.0,
"epoch": 0.4416,
"grad_norm": 0.00026661829906515777,
"learning_rate": 7.258064516129033e-07,
"loss": 0.0082,
"num_tokens": 80884856.0,
"reward": 0.806640625,
"reward_std": 0.10683833807706833,
"rewards/accuracy_reward": 0.6256510615348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9876301884651184,
"rewards/mean_confidence_reward": 0.0,
"step": 69
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 2801.0,
"completions/mean_length": 655.6484375,
"completions/mean_terminated_length": 639.89794921875,
"completions/min_length": 127.0,
"completions/min_terminated_length": 127.0,
"epoch": 0.448,
"grad_norm": 0.0003659256035462022,
"learning_rate": 6.451612903225807e-07,
"loss": 0.0045,
"num_tokens": 82146524.0,
"reward": 0.8385416865348816,
"reward_std": 0.11747505515813828,
"rewards/accuracy_reward": 0.681640625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9954426884651184,
"rewards/mean_confidence_reward": 0.0,
"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": 0.01171875,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3034.0,
"completions/mean_length": 659.3053588867188,
"completions/mean_terminated_length": 618.5540161132812,
"completions/min_length": 160.0,
"completions/min_terminated_length": 160.0,
"epoch": 0.4544,
"grad_norm": 0.0002962271682918072,
"learning_rate": 5.645161290322581e-07,
"loss": 0.0081,
"num_tokens": 83408017.0,
"reward": 0.8626302480697632,
"reward_std": 0.10841435194015503,
"rewards/accuracy_reward": 0.7369791865348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.98828125,
"rewards/mean_confidence_reward": 0.0,
"step": 71
},
{
"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": 0.00455729166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4015.0,
"completions/mean_length": 649.5560302734375,
"completions/mean_terminated_length": 633.777587890625,
"completions/min_length": 149.0,
"completions/min_terminated_length": 149.0,
"epoch": 0.4608,
"grad_norm": 0.00029511775937862694,
"learning_rate": 4.838709677419355e-07,
"loss": 0.0044,
"num_tokens": 84660903.0,
"reward": 0.8092448115348816,
"reward_std": 0.12138192355632782,
"rewards/accuracy_reward": 0.6243489384651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.994140625,
"rewards/mean_confidence_reward": 0.0,
"step": 72
},
{
"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": 0.00846354166666663,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3994.0,
"completions/mean_length": 662.8704833984375,
"completions/mean_terminated_length": 633.5659790039062,
"completions/min_length": 102.0,
"completions/min_terminated_length": 102.0,
"epoch": 0.4672,
"grad_norm": 0.00026064313715323806,
"learning_rate": 4.032258064516129e-07,
"loss": 0.0041,
"num_tokens": 85924768.0,
"reward": 0.8209635615348816,
"reward_std": 0.10173972696065903,
"rewards/accuracy_reward": 0.650390625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9915364384651184,
"rewards/mean_confidence_reward": 0.0,
"step": 73
},
{
"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": 0.001953125,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4057.0,
"completions/mean_length": 613.3997802734375,
"completions/mean_terminated_length": 606.58447265625,
"completions/min_length": 125.0,
"completions/min_terminated_length": 125.0,
"epoch": 0.4736,
"grad_norm": 0.00026255525881424546,
"learning_rate": 3.2258064516129035e-07,
"loss": 0.0037,
"num_tokens": 87117086.0,
"reward": 0.8444010615348816,
"reward_std": 0.08728945255279541,
"rewards/accuracy_reward": 0.69140625,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9973958134651184,
"rewards/mean_confidence_reward": 0.0,
"step": 74
},
{
"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": 0.00911458333333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 4059.0,
"completions/mean_length": 667.43359375,
"completions/mean_terminated_length": 635.8961791992188,
"completions/min_length": 160.0,
"completions/min_terminated_length": 160.0,
"epoch": 0.48,
"grad_norm": 0.0002573035017121583,
"learning_rate": 2.4193548387096775e-07,
"loss": 0.0061,
"num_tokens": 88391336.0,
"reward": 0.8186849355697632,
"reward_std": 0.11762097477912903,
"rewards/accuracy_reward": 0.6471354365348816,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.990234375,
"rewards/mean_confidence_reward": 0.0,
"step": 75
},
{
"epoch": 0.48,
"eval_completions/clipped_ratio": 0.0048828125,
"eval_completions/max_length": 2728.25,
"eval_completions/max_terminated_length": 2279.75,
"eval_completions/mean_length": 641.2385101318359,
"eval_completions/mean_terminated_length": 624.3703231811523,
"eval_completions/min_length": 189.125,
"eval_completions/min_terminated_length": 189.125,
"eval_loss": 0.0,
"eval_num_tokens": 88391336.0,
"eval_reward": 0.8330078125,
"eval_reward_std": 0.23016384057700634,
"eval_rewards/accuracy_reward": 0.6708984375,
"eval_rewards/brier_reward": 0.0,
"eval_rewards/confidence_one_or_zero": 0.0,
"eval_rewards/format_reward": 0.9951171875,
"eval_rewards/mean_confidence_reward": 0.0,
"eval_runtime": 182.2902,
"eval_samples_per_second": 5.486,
"eval_steps_per_second": 0.044,
"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": 0.0032552083333333703,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3493.0,
"completions/mean_length": 640.0716552734375,
"completions/mean_terminated_length": 628.7850952148438,
"completions/min_length": 120.0,
"completions/min_terminated_length": 120.0,
"epoch": 0.4864,
"grad_norm": 0.00025482551427558064,
"learning_rate": 1.6129032258064518e-07,
"loss": 0.0043,
"num_tokens": 89623294.0,
"reward": 0.8557943105697632,
"reward_std": 0.11010631918907166,
"rewards/accuracy_reward": 0.71484375,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9967448115348816,
"rewards/mean_confidence_reward": 0.0,
"step": 76
},
{
"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": 0.00716145833333337,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3253.0,
"completions/mean_length": 644.4609375,
"completions/mean_terminated_length": 619.5645751953125,
"completions/min_length": 115.0,
"completions/min_terminated_length": 115.0,
"epoch": 0.4928,
"grad_norm": 0.00024977046996355057,
"learning_rate": 8.064516129032259e-08,
"loss": 0.0053,
"num_tokens": 90863114.0,
"reward": 0.8515625,
"reward_std": 0.09298877418041229,
"rewards/accuracy_reward": 0.7102864384651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.9928385615348816,
"rewards/mean_confidence_reward": 0.0,
"step": 77
},
{
"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": 0.00390625,
"completions/max_length": 4096.0,
"completions/max_terminated_length": 3669.0,
"completions/mean_length": 647.9694213867188,
"completions/mean_terminated_length": 634.44775390625,
"completions/min_length": 151.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.4992,
"grad_norm": 0.0002479875402059406,
"learning_rate": 0.0,
"loss": 0.0024,
"num_tokens": 92113427.0,
"reward": 0.8307291865348816,
"reward_std": 0.11155258119106293,
"rewards/accuracy_reward": 0.6653645634651184,
"rewards/brier_reward": 0.0,
"rewards/confidence_one_or_zero": 0.0,
"rewards/format_reward": 0.99609375,
"rewards/mean_confidence_reward": 0.0,
"step": 78
},
{
"epoch": 0.4992,
"step": 78,
"total_flos": 0.0,
"train_loss": 0.00840686316578052,
"train_runtime": 15684.0191,
"train_samples_per_second": 0.956,
"train_steps_per_second": 0.005
}
],
"logging_steps": 1,
"max_steps": 78,
"num_input_tokens_seen": 92113427,
"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": 3,
"trial_name": null,
"trial_params": null
}