trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 48,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.020833333333333332,
      "grad_norm": 8.363889637744881,
      "learning_rate": 1e-07,
      "logits/chosen": -2.8258156776428223,
      "logits/rejected": -2.859372615814209,
      "logps/chosen": -247.59397888183594,
      "logps/pi_response": -72.04409790039062,
      "logps/ref_response": -72.04409790039062,
      "logps/rejected": -180.8425750732422,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 9.068207013086154,
      "learning_rate": 4.83504027183137e-07,
      "logits/chosen": -2.7512786388397217,
      "logits/rejected": -2.7331552505493164,
      "logps/chosen": -219.46205139160156,
      "logps/pi_response": -65.90792083740234,
      "logps/ref_response": -65.22713470458984,
      "logps/rejected": -164.32415771484375,
      "loss": 0.6879,
      "rewards/accuracies": 0.5833333134651184,
      "rewards/chosen": 0.015046126209199429,
      "rewards/margins": 0.013462345115840435,
      "rewards/rejected": 0.001583782723173499,
      "step": 10
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 6.854500780501718,
      "learning_rate": 3.643105808261596e-07,
      "logits/chosen": -2.7228639125823975,
      "logits/rejected": -2.692481756210327,
      "logps/chosen": -250.13565063476562,
      "logps/pi_response": -79.09022521972656,
      "logps/ref_response": -66.51771545410156,
      "logps/rejected": -163.60781860351562,
      "loss": 0.6553,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": 0.031493671238422394,
      "rewards/margins": 0.09395480155944824,
      "rewards/rejected": -0.062461137771606445,
      "step": 20
    },
    {
      "epoch": 0.625,
      "grad_norm": 6.610726899137987,
      "learning_rate": 1.8676665440207977e-07,
      "logits/chosen": -2.642488956451416,
      "logits/rejected": -2.608694076538086,
      "logps/chosen": -258.0551452636719,
      "logps/pi_response": -108.9342041015625,
      "logps/ref_response": -70.92167663574219,
      "logps/rejected": -199.82693481445312,
      "loss": 0.6245,
      "rewards/accuracies": 0.7281249761581421,
      "rewards/chosen": -0.08083241432905197,
      "rewards/margins": 0.20813515782356262,
      "rewards/rejected": -0.2889675498008728,
      "step": 30
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 6.863576606705015,
      "learning_rate": 4.1500545527530544e-08,
      "logits/chosen": -2.6711089611053467,
      "logits/rejected": -2.641400098800659,
      "logps/chosen": -272.2811584472656,
      "logps/pi_response": -128.9022216796875,
      "logps/ref_response": -76.34043884277344,
      "logps/rejected": -220.0448455810547,
      "loss": 0.6048,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": -0.16047780215740204,
      "rewards/margins": 0.25879645347595215,
      "rewards/rejected": -0.419274240732193,
      "step": 40
    },
    {
      "epoch": 1.0,
      "step": 48,
      "total_flos": 0.0,
      "train_loss": 0.6321735282739004,
      "train_runtime": 2185.2623,
      "train_samples_per_second": 5.595,
      "train_steps_per_second": 0.022
    }
  ],
  "logging_steps": 10,
  "max_steps": 48,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}