{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.00026,
  "eval_steps": 500,
  "global_step": 13,
  "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": 0.0,
      "completions/max_length": 4368.0,
      "completions/max_terminated_length": 4368.0,
      "completions/mean_length": 3872.46875,
      "completions/mean_terminated_length": 3872.46875,
      "completions/min_length": 1039.0,
      "completions/min_terminated_length": 1039.0,
      "entropy": 0.5444511137902737,
      "epoch": 2e-05,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.4791297912597656,
      "kl": 0.0,
      "learning_rate": 0.0,
      "loss": 0.0766,
      "num_tokens": 147182.0,
      "reward": -2.4586520195007324,
      "reward_std": 1.220275640487671,
      "rewards/rollout_reward_func/mean": -2.4586520195007324,
      "rewards/rollout_reward_func/std": 1.256102204322815,
      "sampling/importance_sampling_ratio/max": 1.6867375373840332,
      "sampling/importance_sampling_ratio/mean": 0.9737950563430786,
      "sampling/importance_sampling_ratio/min": 0.27162742614746094,
      "sampling/sampling_logp_difference/max": 0.7351722717285156,
      "sampling/sampling_logp_difference/mean": 0.02089117467403412,
      "step": 1,
      "step_time": 57.306324382999264
    },
    {
      "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,
      "entropy": 0.5444511137902737,
      "epoch": 4e-05,
      "grad_norm": 3.456141710281372,
      "kl": 0.0,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 0.0766,
      "step": 2,
      "step_time": 11.957301280001047
    },
    {
      "clip_ratio/high_max": 0.0016891892300918698,
      "clip_ratio/high_mean": 0.0008445946150459349,
      "clip_ratio/low_mean": 0.0026041667442768812,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.003448761359322816,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 4214.0,
      "completions/max_terminated_length": 4214.0,
      "completions/mean_length": 3857.65625,
      "completions/mean_terminated_length": 3857.65625,
      "completions/min_length": 2976.0,
      "completions/min_terminated_length": 2976.0,
      "entropy": 0.5738473497331142,
      "epoch": 6e-05,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.3130648136138916,
      "kl": 0.0009115237262449227,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.0672,
      "num_tokens": 294237.0,
      "reward": -2.499704122543335,
      "reward_std": 1.0055248737335205,
      "rewards/rollout_reward_func/mean": -2.499704122543335,
      "rewards/rollout_reward_func/std": 1.122406005859375,
      "sampling/importance_sampling_ratio/max": 1.5998320579528809,
      "sampling/importance_sampling_ratio/mean": 0.9918304085731506,
      "sampling/importance_sampling_ratio/min": 0.6382073163986206,
      "sampling/sampling_logp_difference/max": 0.33383655548095703,
      "sampling/sampling_logp_difference/mean": 0.021819621324539185,
      "step": 3,
      "step_time": 55.93842194800072
    },
    {
      "clip_ratio/high_max": 0.004406580585055053,
      "clip_ratio/high_mean": 0.0022032902925275266,
      "clip_ratio/low_mean": 0.005063657532446086,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.007266947708558291,
      "entropy": 0.5751528590917587,
      "epoch": 8e-05,
      "grad_norm": 2.254608631134033,
      "kl": 0.0011329303233651444,
      "learning_rate": 8.571428571428572e-07,
      "loss": 0.065,
      "step": 4,
      "step_time": 11.812342160998924
    },
    {
      "clip_ratio/high_max": 0.007812500232830644,
      "clip_ratio/high_mean": 0.003906250116415322,
      "clip_ratio/low_mean": 0.0026041667442768812,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.006510416860692203,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 4288.0,
      "completions/max_terminated_length": 4288.0,
      "completions/mean_length": 3989.28125,
      "completions/mean_terminated_length": 3989.28125,
      "completions/min_length": 2506.0,
      "completions/min_terminated_length": 2506.0,
      "entropy": 0.5535362251102924,
      "epoch": 0.0001,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.2777466773986816,
      "kl": 0.0011825040710391477,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.0929,
      "num_tokens": 445258.0,
      "reward": -2.45902681350708,
      "reward_std": 0.6927163600921631,
      "rewards/rollout_reward_func/mean": -2.45902681350708,
      "rewards/rollout_reward_func/std": 0.8290045261383057,
      "sampling/importance_sampling_ratio/max": 1.6969612836837769,
      "sampling/importance_sampling_ratio/mean": 1.0004469156265259,
      "sampling/importance_sampling_ratio/min": 0.5009263157844543,
      "sampling/sampling_logp_difference/max": 0.36035799980163574,
      "sampling/sampling_logp_difference/mean": 0.02125256508588791,
      "step": 5,
      "step_time": 60.257389979000436
    },
    {
      "clip_ratio/high_max": 0.0026041667442768812,
      "clip_ratio/high_mean": 0.0013020833721384406,
      "clip_ratio/low_mean": 0.0013586956774815917,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0026607790496200323,
      "entropy": 0.5545530468225479,
      "epoch": 0.00012,
      "grad_norm": 2.07369065284729,
      "kl": 0.0011546705500222743,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.0921,
      "step": 6,
      "step_time": 11.93783174300097
    },
    {
      "clip_ratio/high_max": 0.0026041667442768812,
      "clip_ratio/high_mean": 0.0013020833721384406,
      "clip_ratio/low_mean": 0.0013020833721384406,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0026041667442768812,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 4363.0,
      "completions/max_terminated_length": 4363.0,
      "completions/mean_length": 3765.53125,
      "completions/mean_terminated_length": 3765.53125,
      "completions/min_length": 392.0,
      "completions/min_terminated_length": 392.0,
      "entropy": 0.573919378221035,
      "epoch": 0.00014,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.3299458026885986,
      "kl": 0.0011782783294620458,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.0167,
      "num_tokens": 589832.0,
      "reward": -2.46976375579834,
      "reward_std": 1.2192020416259766,
      "rewards/rollout_reward_func/mean": -2.46976375579834,
      "rewards/rollout_reward_func/std": 1.360211730003357,
      "sampling/importance_sampling_ratio/max": 1.4216840267181396,
      "sampling/importance_sampling_ratio/mean": 0.9407027959823608,
      "sampling/importance_sampling_ratio/min": 0.5079315900802612,
      "sampling/sampling_logp_difference/max": 0.42226099967956543,
      "sampling/sampling_logp_difference/mean": 0.022609494626522064,
      "step": 7,
      "step_time": 54.58579348800049
    },
    {
      "clip_ratio/high_max": 0.007015306269749999,
      "clip_ratio/high_mean": 0.005070153274573386,
      "clip_ratio/low_mean": 0.0013020833721384406,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.006372236646711826,
      "entropy": 0.5752841830253601,
      "epoch": 0.00016,
      "grad_norm": 2.360698938369751,
      "kl": 0.001157285601948388,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0157,
      "step": 8,
      "step_time": 12.033116390001851
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0026041667442768812,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0026041667442768812,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 4305.0,
      "completions/max_terminated_length": 4305.0,
      "completions/mean_length": 3619.4375,
      "completions/mean_terminated_length": 3619.4375,
      "completions/min_length": 1128.0,
      "completions/min_terminated_length": 1128.0,
      "entropy": 0.5393632538616657,
      "epoch": 0.00018,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.6401772499084473,
      "kl": 0.0009530504758004099,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.0927,
      "num_tokens": 729170.0,
      "reward": -2.326364517211914,
      "reward_std": 1.2151319980621338,
      "rewards/rollout_reward_func/mean": -2.326364517211914,
      "rewards/rollout_reward_func/std": 1.404052972793579,
      "sampling/importance_sampling_ratio/max": 1.7163654565811157,
      "sampling/importance_sampling_ratio/mean": 0.9872620105743408,
      "sampling/importance_sampling_ratio/min": 0.5921608805656433,
      "sampling/sampling_logp_difference/max": 0.5117590427398682,
      "sampling/sampling_logp_difference/mean": 0.024541478604078293,
      "step": 9,
      "step_time": 53.8118479720024
    },
    {
      "clip_ratio/high_max": 0.0026041667442768812,
      "clip_ratio/high_mean": 0.0013020833721384406,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0013020833721384406,
      "entropy": 0.5393857695162296,
      "epoch": 0.0002,
      "grad_norm": 2.6921041011810303,
      "kl": 0.0009762203881109599,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.0925,
      "step": 10,
      "step_time": 12.179582908999691
    },
    {
      "clip_ratio/high_max": 0.0026041667442768812,
      "clip_ratio/high_mean": 0.0013020833721384406,
      "clip_ratio/low_mean": 0.0013020833721384406,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0026041667442768812,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 4316.0,
      "completions/max_terminated_length": 4316.0,
      "completions/mean_length": 3972.75,
      "completions/mean_terminated_length": 3972.75,
      "completions/min_length": 2992.0,
      "completions/min_terminated_length": 2992.0,
      "entropy": 0.5662294253706932,
      "epoch": 0.00022,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.0718979835510254,
      "kl": 0.0011709001882991288,
      "learning_rate": 2.8571428571428573e-06,
      "loss": -0.0709,
      "num_tokens": 879951.0,
      "reward": -2.6862502098083496,
      "reward_std": 0.7006374597549438,
      "rewards/rollout_reward_func/mean": -2.6862502098083496,
      "rewards/rollout_reward_func/std": 0.7967818379402161,
      "sampling/importance_sampling_ratio/max": 1.4347306489944458,
      "sampling/importance_sampling_ratio/mean": 0.9685467481613159,
      "sampling/importance_sampling_ratio/min": 0.6337878704071045,
      "sampling/sampling_logp_difference/max": 0.2577688694000244,
      "sampling/sampling_logp_difference/mean": 0.02011028863489628,
      "step": 11,
      "step_time": 56.197459413000615
    },
    {
      "clip_ratio/high_max": 0.005055147223174572,
      "clip_ratio/high_mean": 0.002527573611587286,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.002527573611587286,
      "entropy": 0.5676523223519325,
      "epoch": 0.00024,
      "grad_norm": 2.133882522583008,
      "kl": 0.0011278250749455765,
      "learning_rate": 3.142857142857143e-06,
      "loss": -0.0707,
      "step": 12,
      "step_time": 12.02025387499998
    },
    {
      "clip_ratio/high_max": 0.0026041667442768812,
      "clip_ratio/high_mean": 0.0013020833721384406,
      "clip_ratio/low_mean": 0.0013020833721384406,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0026041667442768812,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 4299.0,
      "completions/max_terminated_length": 4299.0,
      "completions/mean_length": 3730.71875,
      "completions/mean_terminated_length": 3730.71875,
      "completions/min_length": 1093.0,
      "completions/min_terminated_length": 1093.0,
      "entropy": 0.5347183495759964,
      "epoch": 0.00026,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.1146907806396484,
      "kl": 0.0010681173953344114,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.1176,
      "num_tokens": 1023348.0,
      "reward": -2.0156006813049316,
      "reward_std": 1.2150049209594727,
      "rewards/rollout_reward_func/mean": -2.0156006813049316,
      "rewards/rollout_reward_func/std": 1.5709649324417114,
      "sampling/importance_sampling_ratio/max": 1.7714853286743164,
      "sampling/importance_sampling_ratio/mean": 1.0881149768829346,
      "sampling/importance_sampling_ratio/min": 0.5891880989074707,
      "sampling/sampling_logp_difference/max": 0.34902477264404297,
      "sampling/sampling_logp_difference/mean": 0.018774278461933136,
      "step": 13,
      "step_time": 55.21850023600109
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 200000,
  "num_input_tokens_seen": 1023348,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}