trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.985781990521327,
  "eval_steps": 100,
  "global_step": 52,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.018957345971563982,
      "grad_norm": 3.021558692298453,
      "learning_rate": 8.333333333333333e-07,
      "logits/chosen": -0.5616407990455627,
      "logits/rejected": -0.42516714334487915,
      "logps/chosen": -373.83282470703125,
      "logps/rejected": -493.9554748535156,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.1895734597156398,
      "grad_norm": 3.23447571449701,
      "learning_rate": 4.907293218369499e-06,
      "logits/chosen": -0.1685400903224945,
      "logits/rejected": -0.19170813262462616,
      "logps/chosen": -464.29693603515625,
      "logps/rejected": -501.8587341308594,
      "loss": 0.6936,
      "rewards/accuracies": 0.4097222089767456,
      "rewards/chosen": 0.0008683237829245627,
      "rewards/margins": 0.0007554867188446224,
      "rewards/rejected": 0.00011283738422207534,
      "step": 10
    },
    {
      "epoch": 0.3791469194312796,
      "grad_norm": 3.9389946840001264,
      "learning_rate": 3.941700805287169e-06,
      "logits/chosen": -0.27595698833465576,
      "logits/rejected": -0.21783480048179626,
      "logps/chosen": -438.6907653808594,
      "logps/rejected": -475.0658264160156,
      "loss": 0.6894,
      "rewards/accuracies": 0.4937500059604645,
      "rewards/chosen": -0.003008360043168068,
      "rewards/margins": 0.0011918289819732308,
      "rewards/rejected": -0.00420018844306469,
      "step": 20
    },
    {
      "epoch": 0.5687203791469194,
      "grad_norm": 3.2779169381175386,
      "learning_rate": 2.3293939665883233e-06,
      "logits/chosen": -0.012437907047569752,
      "logits/rejected": -0.20707044005393982,
      "logps/chosen": -476.6265563964844,
      "logps/rejected": -474.9481506347656,
      "loss": 0.6873,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.01446505170315504,
      "rewards/margins": 0.007696834392845631,
      "rewards/rejected": -0.02216188609600067,
      "step": 30
    },
    {
      "epoch": 0.7582938388625592,
      "grad_norm": 2.8592366061141954,
      "learning_rate": 7.936171419533653e-07,
      "logits/chosen": -0.4586310386657715,
      "logits/rejected": -0.5929983854293823,
      "logps/chosen": -422.44293212890625,
      "logps/rejected": -402.6429748535156,
      "loss": 0.6827,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -0.018473980948328972,
      "rewards/margins": 0.014362658374011517,
      "rewards/rejected": -0.03283664211630821,
      "step": 40
    },
    {
      "epoch": 0.9478672985781991,
      "grad_norm": 3.296382116125289,
      "learning_rate": 2.3285134909173113e-08,
      "logits/chosen": 0.08067023754119873,
      "logits/rejected": 0.19673587381839752,
      "logps/chosen": -483.49212646484375,
      "logps/rejected": -536.2080688476562,
      "loss": 0.679,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": -0.017066333442926407,
      "rewards/margins": 0.01992700807750225,
      "rewards/rejected": -0.03699334338307381,
      "step": 50
    },
    {
      "epoch": 0.985781990521327,
      "step": 52,
      "total_flos": 0.0,
      "train_loss": 0.6860739244864538,
      "train_runtime": 316.322,
      "train_samples_per_second": 21.339,
      "train_steps_per_second": 0.164
    }
  ],
  "logging_steps": 10,
  "max_steps": 52,
  "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": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}