trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9963570127504555,
  "eval_steps": 500,
  "global_step": 548,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.18214936247723132,
      "grad_norm": 2.3424386978149414,
      "learning_rate": 4.5454545454545455e-06,
      "logits/chosen": 0.3313373327255249,
      "logits/rejected": 0.51214599609375,
      "logps/chosen": -133.43138122558594,
      "logps/rejected": -192.70355224609375,
      "loss": 0.4056,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.41640931367874146,
      "rewards/margins": 0.7543150186538696,
      "rewards/rejected": -0.33790573477745056,
      "step": 50
    },
    {
      "epoch": 0.36429872495446264,
      "grad_norm": 2.2886452674865723,
      "learning_rate": 4.897914773180057e-06,
      "logits/chosen": 0.03562241047620773,
      "logits/rejected": 0.31711575388908386,
      "logps/chosen": -147.89747619628906,
      "logps/rejected": -268.35821533203125,
      "loss": 0.2839,
      "rewards/accuracies": 0.9950000047683716,
      "rewards/chosen": 0.28251099586486816,
      "rewards/margins": 1.3375725746154785,
      "rewards/rejected": -1.0550616979599,
      "step": 100
    },
    {
      "epoch": 0.546448087431694,
      "grad_norm": 3.107114791870117,
      "learning_rate": 4.5557177668257975e-06,
      "logits/chosen": -0.2743435502052307,
      "logits/rejected": -0.10878603160381317,
      "logps/chosen": -208.27593994140625,
      "logps/rejected": -473.1844787597656,
      "loss": 0.157,
      "rewards/accuracies": 1.0,
      "rewards/chosen": -0.3748534321784973,
      "rewards/margins": 2.790405511856079,
      "rewards/rejected": -3.1652591228485107,
      "step": 150
    },
    {
      "epoch": 0.7285974499089253,
      "grad_norm": 6.934810638427734,
      "learning_rate": 4.006586590948141e-06,
      "logits/chosen": -0.08417636901140213,
      "logits/rejected": -0.03048125095665455,
      "logps/chosen": -362.09002685546875,
      "logps/rejected": -824.4659423828125,
      "loss": 0.0918,
      "rewards/accuracies": 0.9850000143051147,
      "rewards/chosen": -1.878377079963684,
      "rewards/margins": 4.862216949462891,
      "rewards/rejected": -6.740594387054443,
      "step": 200
    },
    {
      "epoch": 0.9107468123861566,
      "grad_norm": 7.77178430557251,
      "learning_rate": 3.3057982907103e-06,
      "logits/chosen": 0.220669686794281,
      "logits/rejected": 0.25975489616394043,
      "logps/chosen": -471.4226989746094,
      "logps/rejected": -1041.406005859375,
      "loss": 0.1067,
      "rewards/accuracies": 0.9850000143051147,
      "rewards/chosen": -2.964336633682251,
      "rewards/margins": 5.914289474487305,
      "rewards/rejected": -8.878626823425293,
      "step": 250
    },
    {
      "epoch": 1.092896174863388,
      "grad_norm": 0.5947583317756653,
      "learning_rate": 2.5238961319543297e-06,
      "logits/chosen": 0.16262899339199066,
      "logits/rejected": 0.25519824028015137,
      "logps/chosen": -494.9023132324219,
      "logps/rejected": -1173.1524658203125,
      "loss": 0.0808,
      "rewards/accuracies": 0.9850000143051147,
      "rewards/chosen": -3.2024941444396973,
      "rewards/margins": 6.978786468505859,
      "rewards/rejected": -10.181281089782715,
      "step": 300
    },
    {
      "epoch": 1.2750455373406193,
      "grad_norm": 3.096635580062866,
      "learning_rate": 1.7395885232379676e-06,
      "logits/chosen": -0.01755141280591488,
      "logits/rejected": 0.047465089708566666,
      "logps/chosen": -570.8333740234375,
      "logps/rejected": -1329.80615234375,
      "loss": 0.0538,
      "rewards/accuracies": 0.9950000047683716,
      "rewards/chosen": -3.900271415710449,
      "rewards/margins": 7.754390716552734,
      "rewards/rejected": -11.654662132263184,
      "step": 350
    },
    {
      "epoch": 1.4571948998178508,
      "grad_norm": 1.1082794666290283,
      "learning_rate": 1.0318260122870355e-06,
      "logits/chosen": -0.10038640350103378,
      "logits/rejected": -0.03151293843984604,
      "logps/chosen": -551.974609375,
      "logps/rejected": -1289.09375,
      "loss": 0.0262,
      "rewards/accuracies": 1.0,
      "rewards/chosen": -3.812049150466919,
      "rewards/margins": 7.496457099914551,
      "rewards/rejected": -11.30850601196289,
      "step": 400
    },
    {
      "epoch": 1.639344262295082,
      "grad_norm": 5.007543087005615,
      "learning_rate": 4.718539080573295e-07,
      "logits/chosen": -0.21412551403045654,
      "logits/rejected": -0.1368548423051834,
      "logps/chosen": -588.555419921875,
      "logps/rejected": -1397.4381103515625,
      "loss": 0.0756,
      "rewards/accuracies": 0.9800000190734863,
      "rewards/chosen": -4.129397392272949,
      "rewards/margins": 8.312859535217285,
      "rewards/rejected": -12.442255973815918,
      "step": 450
    },
    {
      "epoch": 1.8214936247723132,
      "grad_norm": 0.07571525871753693,
      "learning_rate": 1.1604053319265923e-07,
      "logits/chosen": -0.27741381525993347,
      "logits/rejected": -0.19134016335010529,
      "logps/chosen": -600.4368896484375,
      "logps/rejected": -1366.5653076171875,
      "loss": 0.0695,
      "rewards/accuracies": 0.9850000143051147,
      "rewards/chosen": -4.229435920715332,
      "rewards/margins": 7.8976263999938965,
      "rewards/rejected": -12.12706184387207,
      "step": 500
    }
  ],
  "logging_steps": 50,
  "max_steps": 548,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}