Qwen3-Codeforces-GRPO / trainer_state.json
yarin-shaked's picture
Model save
2b23d24 verified
Invalid JSON: Unexpected token 'N', ..."igh_max": NaN, "... is not valid JSON
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 32,
"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": -0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 8050.0,
"completions/mean_length": 6862.1875,
"completions/mean_terminated_length": 3936.60009765625,
"completions/min_length": 1679.0,
"completions/min_terminated_length": 1679.0,
"epoch": 0.125,
"frac_reward_zero_std": 0.5,
"grad_norm": 0.1540709645657331,
"learning_rate": 0.0,
"loss": -0.0397,
"num_tokens": 31407.0,
"reward": -0.699999988079071,
"reward_std": 0.125,
"rewards/cf_code_reward/mean": -0.699999988079071,
"rewards/cf_code_reward/std": 0.46043458580970764,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 1
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.25,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 42723.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.75,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7057.0,
"completions/mean_length": 8121.0625,
"completions/mean_terminated_length": 7057.0,
"completions/min_length": 7057.0,
"completions/min_terminated_length": 7057.0,
"epoch": 0.375,
"frac_reward_zero_std": 0.75,
"grad_norm": 0.16390367065221123,
"learning_rate": 1e-06,
"loss": -0.0404,
"num_tokens": 62168.0,
"reward": -0.9375,
"reward_std": 0.125,
"rewards/cf_code_reward/mean": -0.9375,
"rewards/cf_code_reward/std": 0.25,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7780.0,
"completions/mean_length": 7816.625,
"completions/mean_terminated_length": 6190.0,
"completions/min_length": 4315.0,
"completions/min_terminated_length": 4315.0,
"epoch": 0.5,
"frac_reward_zero_std": 0.75,
"grad_norm": 0.2878601922045561,
"learning_rate": 1e-06,
"loss": -0.0875,
"num_tokens": 91526.0,
"reward": -0.8125,
"reward_std": 0.248746857047081,
"rewards/cf_code_reward/mean": -0.8125,
"rewards/cf_code_reward/std": 0.5572252869606018,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": -0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 6331.0,
"completions/mean_length": 6324.3125,
"completions/mean_terminated_length": 2215.400146484375,
"completions/min_length": 905.0,
"completions/min_terminated_length": 905.0,
"epoch": 0.625,
"frac_reward_zero_std": 0.5,
"grad_norm": 0.24502748244021016,
"learning_rate": 1e-06,
"loss": -0.0402,
"num_tokens": 118167.0,
"reward": -0.6746212244033813,
"reward_std": 0.428866446018219,
"rewards/cf_code_reward/mean": -0.6746212244033813,
"rewards/cf_code_reward/std": 0.7365784049034119,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": -1.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 5359.0,
"completions/mean_length": 5666.6875,
"completions/mean_terminated_length": 3702.5556640625,
"completions/min_length": 1851.0,
"completions/min_terminated_length": 1851.0,
"epoch": 0.75,
"frac_reward_zero_std": 0.25,
"grad_norm": 0.21413429634446854,
"learning_rate": 1e-06,
"loss": -0.0416,
"num_tokens": 161958.0,
"reward": -0.5037499666213989,
"reward_std": 0.39220669865608215,
"rewards/cf_code_reward/mean": -0.5037499666213989,
"rewards/cf_code_reward/std": 0.6423693299293518,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.75,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7498.0,
"completions/mean_length": 8148.625,
"completions/mean_terminated_length": 7498.0,
"completions/min_length": 7498.0,
"completions/min_terminated_length": 7498.0,
"epoch": 0.875,
"frac_reward_zero_std": 0.75,
"grad_norm": 0.15542675545894089,
"learning_rate": 1e-06,
"loss": -0.0386,
"num_tokens": 182668.0,
"reward": -0.9437500238418579,
"reward_std": 0.11249999701976776,
"rewards/cf_code_reward/mean": -0.9437500238418579,
"rewards/cf_code_reward/std": 0.22499999403953552,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7341.0,
"completions/mean_length": 7525.6875,
"completions/mean_terminated_length": 4638.33349609375,
"completions/min_length": 2906.0,
"completions/min_terminated_length": 2906.0,
"epoch": 1.0,
"frac_reward_zero_std": 0.5,
"grad_norm": 0.17998151407614052,
"learning_rate": 1e-06,
"loss": -0.0646,
"num_tokens": 209383.0,
"reward": -0.8104166984558105,
"reward_std": 0.2664600610733032,
"rewards/cf_code_reward/mean": -0.8104166984558105,
"rewards/cf_code_reward/std": 0.4091035723686218,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7326.0,
"completions/mean_length": 7806.75,
"completions/mean_terminated_length": 6137.33349609375,
"completions/min_length": 4799.0,
"completions/min_terminated_length": 4799.0,
"epoch": 0.28125,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2669470114797714,
"learning_rate": 1e-06,
"loss": -0.1097,
"num_tokens": 237667.0,
"reward": -0.8180555105209351,
"reward_std": 0.3918065130710602,
"rewards/cf_code_reward/mean": -0.8180555105209351,
"rewards/cf_code_reward/std": 0.39180654287338257,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.75,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7539.0,
"completions/mean_length": 8151.1875,
"completions/mean_terminated_length": 7539.0,
"completions/min_length": 7539.0,
"completions/min_terminated_length": 7539.0,
"epoch": 0.3125,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.20723698042342187,
"learning_rate": 1e-06,
"loss": -0.0539,
"num_tokens": 266246.0,
"reward": -0.9375,
"reward_std": 0.25,
"rewards/cf_code_reward/mean": -0.9375,
"rewards/cf_code_reward/std": 0.25,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 10
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.34375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 275446.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7984.0,
"completions/mean_length": 7683.0,
"completions/mean_terminated_length": 5477.33349609375,
"completions/min_length": 4031.0,
"completions/min_terminated_length": 4031.0,
"epoch": 0.375,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2634279160052341,
"learning_rate": 1e-06,
"loss": -0.0997,
"num_tokens": 301318.0,
"reward": -0.817991316318512,
"reward_std": 0.39194566011428833,
"rewards/cf_code_reward/mean": -0.817991316318512,
"rewards/cf_code_reward/std": 0.39194566011428833,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 12
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.40625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 310774.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 13
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.4375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 319062.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 6720.0,
"completions/mean_length": 7516.6875,
"completions/mean_terminated_length": 4590.33349609375,
"completions/min_length": 2107.0,
"completions/min_terminated_length": 2107.0,
"epoch": 0.46875,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.31248865871350523,
"learning_rate": 1e-06,
"loss": -0.0589,
"num_tokens": 343457.0,
"reward": -0.8062499761581421,
"reward_std": 0.5662375688552856,
"rewards/cf_code_reward/mean": -0.8062499761581421,
"rewards/cf_code_reward/std": 0.5662376284599304,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"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": -1.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 6767.0,
"completions/mean_length": 6238.25,
"completions/mean_terminated_length": 4718.66650390625,
"completions/min_length": 3229.0,
"completions/min_terminated_length": 3229.0,
"epoch": 0.5,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.44442560969386563,
"learning_rate": 1e-06,
"loss": -0.1683,
"num_tokens": 400709.0,
"reward": -0.38749998807907104,
"reward_std": 0.732916533946991,
"rewards/cf_code_reward/mean": -0.38749998807907104,
"rewards/cf_code_reward/std": 0.7329164743423462,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": -3.0,
"completions/max_length": 1425.0,
"completions/max_terminated_length": 1425.0,
"completions/mean_length": 1088.0625,
"completions/mean_terminated_length": 1088.0625,
"completions/min_length": 753.0,
"completions/min_terminated_length": 753.0,
"epoch": 0.53125,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.3425692374871352,
"learning_rate": 1e-06,
"loss": -0.0026,
"num_tokens": 424198.0,
"reward": -0.2874999940395355,
"reward_std": 0.9912113547325134,
"rewards/cf_code_reward/mean": -0.2874999940395355,
"rewards/cf_code_reward/std": 0.991211473941803,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 17
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.5625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 444422.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 6500.0,
"completions/mean_length": 7518.4375,
"completions/mean_terminated_length": 4599.6669921875,
"completions/min_length": 2613.0,
"completions/min_terminated_length": 2613.0,
"epoch": 0.59375,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2247292061728764,
"learning_rate": 1e-06,
"loss": -0.0863,
"num_tokens": 467501.0,
"reward": -0.8107954263687134,
"reward_std": 0.4067820608615875,
"rewards/cf_code_reward/mean": -0.8107954263687134,
"rewards/cf_code_reward/std": 0.4067820906639099,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.75,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 5671.0,
"completions/mean_length": 8034.4375,
"completions/mean_terminated_length": 5671.0,
"completions/min_length": 5671.0,
"completions/min_terminated_length": 5671.0,
"epoch": 0.625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 499844.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": -2.5,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 6850.0,
"completions/mean_length": 4564.25,
"completions/mean_terminated_length": 4046.000244140625,
"completions/min_length": 1360.0,
"completions/min_terminated_length": 1360.0,
"epoch": 0.65625,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.08550363834268984,
"learning_rate": 1e-06,
"loss": -0.055,
"num_tokens": 564616.0,
"reward": -0.13437500596046448,
"reward_std": 0.3423832058906555,
"rewards/cf_code_reward/mean": -0.13437500596046448,
"rewards/cf_code_reward/std": 0.3423832058906555,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": -3.0,
"completions/max_length": 5018.0,
"completions/max_terminated_length": 5018.0,
"completions/mean_length": 3474.8125,
"completions/mean_terminated_length": 3474.8125,
"completions/min_length": 2417.0,
"completions/min_terminated_length": 2417.0,
"epoch": 0.6875,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.28546210458938714,
"learning_rate": 1e-06,
"loss": 0.0211,
"num_tokens": 630053.0,
"reward": 0.7673437595367432,
"reward_std": 0.5111241936683655,
"rewards/cf_code_reward/mean": 0.7673437595367432,
"rewards/cf_code_reward/std": 0.5111241936683655,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 22
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.71875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 641525.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": -1.5,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7788.0,
"completions/mean_length": 6242.4375,
"completions/mean_terminated_length": 5072.7001953125,
"completions/min_length": 2712.0,
"completions/min_terminated_length": 2712.0,
"epoch": 0.75,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2722849336489413,
"learning_rate": 1e-06,
"loss": -0.1274,
"num_tokens": 704684.0,
"reward": -0.5212500095367432,
"reward_std": 0.4962508976459503,
"rewards/cf_code_reward/mean": -0.5212500095367432,
"rewards/cf_code_reward/std": 0.4962509572505951,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 24
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.78125,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 719116.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.25,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 8087.0,
"completions/mean_length": 8041.8125,
"completions/mean_terminated_length": 7391.0,
"completions/min_length": 6356.0,
"completions/min_terminated_length": 6356.0,
"epoch": 0.8125,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.2942398936933423,
"learning_rate": 1e-06,
"loss": -0.1398,
"num_tokens": 752985.0,
"reward": -0.8091517686843872,
"reward_std": 0.4103502035140991,
"rewards/cf_code_reward/mean": -0.8091517686843872,
"rewards/cf_code_reward/std": 0.4103502035140991,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.75,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7620.0,
"completions/mean_length": 8156.25,
"completions/mean_terminated_length": 7620.0,
"completions/min_length": 7620.0,
"completions/min_terminated_length": 7620.0,
"epoch": 0.84375,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.19931649636776574,
"learning_rate": 1e-06,
"loss": -0.0491,
"num_tokens": 771469.0,
"reward": -0.9437500238418579,
"reward_std": 0.22500000894069672,
"rewards/cf_code_reward/mean": -0.9437500238418579,
"rewards/cf_code_reward/std": 0.22499999403953552,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 27
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 787325.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 28
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 0.90625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 801389.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 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": 0.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 7262.0,
"completions/mean_length": 7653.6875,
"completions/mean_terminated_length": 6038.75,
"completions/min_length": 3702.0,
"completions/min_terminated_length": 3702.0,
"epoch": 0.9375,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.32699225578745883,
"learning_rate": 1e-06,
"loss": -0.1449,
"num_tokens": 834456.0,
"reward": -0.737500011920929,
"reward_std": 0.4698108732700348,
"rewards/cf_code_reward/mean": -0.737500011920929,
"rewards/cf_code_reward/std": 0.4698108732700348,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"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": 0.5,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 4795.0,
"completions/mean_length": 7756.4375,
"completions/mean_terminated_length": 4707.5,
"completions/min_length": 4620.0,
"completions/min_terminated_length": 4620.0,
"epoch": 0.96875,
"frac_reward_zero_std": 0.0,
"grad_norm": 0.331766382104726,
"learning_rate": 1e-06,
"loss": -0.0646,
"num_tokens": 854271.0,
"reward": -0.8687499761581421,
"reward_std": 0.5249999165534973,
"rewards/cf_code_reward/mean": -0.8687499761581421,
"rewards/cf_code_reward/std": 0.5250000357627869,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 31
},
{
"clip_ratio/high_max": NaN,
"clip_ratio/high_mean": NaN,
"clip_ratio/low_mean": NaN,
"clip_ratio/low_min": NaN,
"clip_ratio/region_mean": NaN,
"completions/clipped_ratio": 1.0,
"completions/max_length": 8192.0,
"completions/max_terminated_length": 0.0,
"completions/mean_length": 8192.0,
"completions/mean_terminated_length": 0.0,
"completions/min_length": 8192.0,
"completions/min_terminated_length": 0.0,
"epoch": 1.0,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0,
"learning_rate": 1e-06,
"loss": 0.0,
"num_tokens": 872095.0,
"reward": -1.0,
"reward_std": 0.0,
"rewards/cf_code_reward/mean": -1.0,
"rewards/cf_code_reward/std": 0.0,
"rewards/code_format_reward/mean": 0.0,
"rewards/code_format_reward/std": 0.0,
"step": 32
},
{
"epoch": 1.0,
"step": 32,
"total_flos": 0.0,
"train_loss": 0.0,
"train_runtime": 1.5617,
"train_samples_per_second": 10.245,
"train_steps_per_second": 0.64
}
],
"logging_steps": 1,
"max_steps": 1,
"num_input_tokens_seen": 872095,
"num_train_epochs": 1,
"save_steps": 1,
"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": 4,
"trial_name": null,
"trial_params": null
}