trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 25,
  "global_step": 52,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 8.333333333333333e-08,
      "logits/generated": -2.788468599319458,
      "logits/real": -2.8911099433898926,
      "logps/generated": -226.66921997070312,
      "logps/real": -283.6243896484375,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/generated": 0.0,
      "rewards/margins": 0.0,
      "rewards/real": 0.0,
      "step": 1
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.5652173913043473e-07,
      "logits/generated": -2.8515138626098633,
      "logits/real": -2.8768396377563477,
      "logps/generated": -354.09619140625,
      "logps/real": -350.52911376953125,
      "loss": 0.522,
      "rewards/accuracies": 0.7569444179534912,
      "rewards/generated": -0.21337264776229858,
      "rewards/margins": 0.5563015341758728,
      "rewards/real": 0.3429288864135742,
      "step": 10
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.478260869565217e-07,
      "logits/generated": -2.792628049850464,
      "logits/real": -2.7778868675231934,
      "logps/generated": -351.04638671875,
      "logps/real": -327.13482666015625,
      "loss": 0.3011,
      "rewards/accuracies": 0.90625,
      "rewards/generated": -0.7889599800109863,
      "rewards/margins": 1.546514868736267,
      "rewards/real": 0.7575550675392151,
      "step": 20
    },
    {
      "epoch": 0.96,
      "eval_logits/generated": -2.764375686645508,
      "eval_logits/real": -2.7640507221221924,
      "eval_logps/generated": -310.69891357421875,
      "eval_logps/real": -306.61572265625,
      "eval_loss": 0.24416939914226532,
      "eval_rewards/accuracies": 0.9791666865348816,
      "eval_rewards/generated": -0.9850902557373047,
      "eval_rewards/margins": 2.14570689201355,
      "eval_rewards/real": 1.1606166362762451,
      "eval_runtime": 27.6861,
      "eval_samples_per_second": 6.646,
      "eval_steps_per_second": 0.217,
      "step": 25
    },
    {
      "epoch": 1.15,
      "learning_rate": 2.391304347826087e-07,
      "logits/generated": -2.7530007362365723,
      "logits/real": -2.734692096710205,
      "logps/generated": -310.22607421875,
      "logps/real": -306.02044677734375,
      "loss": 0.1788,
      "rewards/accuracies": 0.9312499761581421,
      "rewards/generated": -1.791497826576233,
      "rewards/margins": 3.7750840187072754,
      "rewards/real": 1.9835857152938843,
      "step": 30
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.3043478260869563e-07,
      "logits/generated": -2.7655322551727295,
      "logits/real": -2.776773691177368,
      "logps/generated": -358.19403076171875,
      "logps/real": -309.92767333984375,
      "loss": 0.0384,
      "rewards/accuracies": 0.9937499761581421,
      "rewards/generated": -3.1963627338409424,
      "rewards/margins": 6.872523307800293,
      "rewards/real": 3.6761608123779297,
      "step": 40
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.1739130434782606e-08,
      "logits/generated": -2.7564454078674316,
      "logits/real": -2.7757363319396973,
      "logps/generated": -357.3354797363281,
      "logps/real": -296.8515930175781,
      "loss": 0.0376,
      "rewards/accuracies": 0.987500011920929,
      "rewards/generated": -2.9351892471313477,
      "rewards/margins": 6.2575507164001465,
      "rewards/real": 3.322361707687378,
      "step": 50
    },
    {
      "epoch": 1.92,
      "eval_logits/generated": -2.7557647228240967,
      "eval_logits/real": -2.7546520233154297,
      "eval_logps/generated": -309.8145446777344,
      "eval_logps/real": -304.967041015625,
      "eval_loss": 0.23592980206012726,
      "eval_rewards/accuracies": 0.9791666865348816,
      "eval_rewards/generated": -0.8966498374938965,
      "eval_rewards/margins": 2.2221336364746094,
      "eval_rewards/real": 1.3254839181900024,
      "eval_runtime": 27.8272,
      "eval_samples_per_second": 6.612,
      "eval_steps_per_second": 0.216,
      "step": 50
    },
    {
      "epoch": 2.0,
      "step": 52,
      "total_flos": 0.0,
      "train_loss": 0.2113667087486157,
      "train_runtime": 1162.1581,
      "train_samples_per_second": 2.836,
      "train_steps_per_second": 0.045
    }
  ],
  "logging_steps": 10,
  "max_steps": 52,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}