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": 8.362677556536177,
      "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": 9.049468118004324,
      "learning_rate": 4.83504027183137e-07,
      "logits/chosen": -2.75138783454895,
      "logits/rejected": -2.7332353591918945,
      "logps/chosen": -219.43704223632812,
      "logps/pi_response": -65.90155029296875,
      "logps/ref_response": -65.22713470458984,
      "logps/rejected": -164.3558807373047,
      "loss": 0.6878,
      "rewards/accuracies": 0.5868055820465088,
      "rewards/chosen": 0.015296363271772861,
      "rewards/margins": 0.014029696583747864,
      "rewards/rejected": 0.0012666649417951703,
      "step": 10
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 6.837906491432947,
      "learning_rate": 3.643105808261596e-07,
      "logits/chosen": -2.722661256790161,
      "logits/rejected": -2.6922712326049805,
      "logps/chosen": -250.12081909179688,
      "logps/pi_response": -79.08014678955078,
      "logps/ref_response": -66.51771545410156,
      "logps/rejected": -163.59561157226562,
      "loss": 0.6554,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": 0.03164178133010864,
      "rewards/margins": 0.09398074448108673,
      "rewards/rejected": -0.06233896687626839,
      "step": 20
    },
    {
      "epoch": 0.625,
      "grad_norm": 6.600290233933933,
      "learning_rate": 1.8676665440207977e-07,
      "logits/chosen": -2.642151355743408,
      "logits/rejected": -2.608363389968872,
      "logps/chosen": -258.04754638671875,
      "logps/pi_response": -108.9358139038086,
      "logps/ref_response": -70.92167663574219,
      "logps/rejected": -199.85496520996094,
      "loss": 0.6245,
      "rewards/accuracies": 0.731249988079071,
      "rewards/chosen": -0.08075634390115738,
      "rewards/margins": 0.20849159359931946,
      "rewards/rejected": -0.28924795985221863,
      "step": 30
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 6.86872847356136,
      "learning_rate": 4.1500545527530544e-08,
      "logits/chosen": -2.6706020832061768,
      "logits/rejected": -2.640864849090576,
      "logps/chosen": -272.2802734375,
      "logps/pi_response": -128.96005249023438,
      "logps/ref_response": -76.34043884277344,
      "logps/rejected": -220.1116943359375,
      "loss": 0.6048,
      "rewards/accuracies": 0.7093750238418579,
      "rewards/chosen": -0.16046889126300812,
      "rewards/margins": 0.2594737112522125,
      "rewards/rejected": -0.41994261741638184,
      "step": 40
    },
    {
      "epoch": 1.0,
      "step": 48,
      "total_flos": 0.0,
      "train_loss": 0.6321128209431967,
      "train_runtime": 2177.6416,
      "train_samples_per_second": 5.615,
      "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
}