trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 96,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5e-08,
      "logits/chosen": -2.848330020904541,
      "logits/rejected": -2.834825038909912,
      "logps/chosen": -287.69622802734375,
      "logps/rejected": -255.33815002441406,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.1,
      "learning_rate": 5e-07,
      "logits/chosen": -2.8012733459472656,
      "logits/rejected": -2.753434658050537,
      "logps/chosen": -244.99864196777344,
      "logps/rejected": -200.799560546875,
      "loss": 0.6926,
      "rewards/accuracies": 0.5069444179534912,
      "rewards/chosen": 0.0004034777812194079,
      "rewards/margins": 0.0008078858954831958,
      "rewards/rejected": -0.00040440805605612695,
      "step": 10
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.83504027183137e-07,
      "logits/chosen": -2.7793192863464355,
      "logits/rejected": -2.7794222831726074,
      "logps/chosen": -267.78729248046875,
      "logps/rejected": -250.0585174560547,
      "loss": 0.6825,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.021286360919475555,
      "rewards/margins": 0.027004068717360497,
      "rewards/rejected": -0.00571770966053009,
      "step": 20
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.3619304568594546e-07,
      "logits/chosen": -2.7962687015533447,
      "logits/rejected": -2.7896554470062256,
      "logps/chosen": -278.6320495605469,
      "logps/rejected": -262.569580078125,
      "loss": 0.6661,
      "rewards/accuracies": 0.6812499761581421,
      "rewards/chosen": 0.0014000388327986002,
      "rewards/margins": 0.0615326389670372,
      "rewards/rejected": -0.060132600367069244,
      "step": 30
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.643105808261596e-07,
      "logits/chosen": -2.8134655952453613,
      "logits/rejected": -2.801830768585205,
      "logps/chosen": -290.64794921875,
      "logps/rejected": -259.28704833984375,
      "loss": 0.6455,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -0.10460182279348373,
      "rewards/margins": 0.1050100103020668,
      "rewards/rejected": -0.20961184799671173,
      "step": 40
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.7734280209446865e-07,
      "logits/chosen": -2.7795636653900146,
      "logits/rejected": -2.7554478645324707,
      "logps/chosen": -303.3760986328125,
      "logps/rejected": -283.14141845703125,
      "loss": 0.6247,
      "rewards/accuracies": 0.6625000238418579,
      "rewards/chosen": -0.15783920884132385,
      "rewards/margins": 0.1601353883743286,
      "rewards/rejected": -0.3179745674133301,
      "step": 50
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8676665440207977e-07,
      "logits/chosen": -2.773420810699463,
      "logits/rejected": -2.792323589324951,
      "logps/chosen": -319.9610900878906,
      "logps/rejected": -301.3622741699219,
      "loss": 0.6152,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -0.33203256130218506,
      "rewards/margins": 0.21394209563732147,
      "rewards/rejected": -0.5459746718406677,
      "step": 60
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.0453527111051183e-07,
      "logits/chosen": -2.770045518875122,
      "logits/rejected": -2.764028549194336,
      "logps/chosen": -324.46160888671875,
      "logps/rejected": -320.1623840332031,
      "loss": 0.5972,
      "rewards/accuracies": 0.6937500238418579,
      "rewards/chosen": -0.2613491415977478,
      "rewards/margins": 0.3139723539352417,
      "rewards/rejected": -0.5753214955329895,
      "step": 70
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.1500545527530544e-08,
      "logits/chosen": -2.8060431480407715,
      "logits/rejected": -2.773552417755127,
      "logps/chosen": -326.8184509277344,
      "logps/rejected": -345.7678527832031,
      "loss": 0.589,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.369906485080719,
      "rewards/margins": 0.32024139165878296,
      "rewards/rejected": -0.690147876739502,
      "step": 80
    },
    {
      "epoch": 0.94,
      "learning_rate": 5.981030611018234e-09,
      "logits/chosen": -2.753633499145508,
      "logits/rejected": -2.697878360748291,
      "logps/chosen": -304.6568603515625,
      "logps/rejected": -340.0152893066406,
      "loss": 0.6002,
      "rewards/accuracies": 0.6937500238418579,
      "rewards/chosen": -0.35058873891830444,
      "rewards/margins": 0.34887048602104187,
      "rewards/rejected": -0.6994592547416687,
      "step": 90
    },
    {
      "epoch": 1.0,
      "step": 96,
      "total_flos": 0.0,
      "train_loss": 0.6312613611419996,
      "train_runtime": 1566.6754,
      "train_samples_per_second": 7.804,
      "train_steps_per_second": 0.061
    }
  ],
  "logging_steps": 10,
  "max_steps": 96,
  "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
}