{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9984,
  "eval_steps": 500,
  "global_step": 156,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.32,
      "grad_norm": 1.796875,
      "learning_rate": 4.423334927457198e-05,
      "logits/chosen": 0.23468714952468872,
      "logits/rejected": 0.07762019336223602,
      "logps/chosen": -339.36993408203125,
      "logps/rejected": -304.6798400878906,
      "loss": 0.6127,
      "rewards/accuracies": 0.6162499785423279,
      "rewards/chosen": 0.062432702630758286,
      "rewards/margins": 0.42034706473350525,
      "rewards/rejected": -0.35791435837745667,
      "step": 50
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8125,
      "learning_rate": 1.815842524819793e-05,
      "logits/chosen": 0.25215524435043335,
      "logits/rejected": 0.08403243869543076,
      "logps/chosen": -330.1611633300781,
      "logps/rejected": -307.3507995605469,
      "loss": 0.4757,
      "rewards/accuracies": 0.7912499904632568,
      "rewards/chosen": 0.5325586795806885,
      "rewards/margins": 1.6245554685592651,
      "rewards/rejected": -1.0919969081878662,
      "step": 100
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.25,
      "learning_rate": 2.397392281198729e-07,
      "logits/chosen": 0.28577932715415955,
      "logits/rejected": 0.10868644714355469,
      "logps/chosen": -331.35308837890625,
      "logps/rejected": -307.7894592285156,
      "loss": 0.4708,
      "rewards/accuracies": 0.7475000023841858,
      "rewards/chosen": 0.3710547983646393,
      "rewards/margins": 1.6936246156692505,
      "rewards/rejected": -1.3225700855255127,
      "step": 150
    },
    {
      "epoch": 0.9984,
      "step": 156,
      "total_flos": 6.505677982198137e+18,
      "train_loss": 0.5174369307664725,
      "train_runtime": 20301.4315,
      "train_samples_per_second": 0.493,
      "train_steps_per_second": 0.008
    }
  ],
  "logging_steps": 50,
  "max_steps": 156,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "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": 6.505677982198137e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}