trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9874476987447699,
  "eval_steps": 500,
  "global_step": 59,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 5e-08,
      "logits/chosen": -2.593817710876465,
      "logits/rejected": -2.562267780303955,
      "logps/chosen": -224.3258819580078,
      "logps/pi_response": -158.2955322265625,
      "logps/ref_response": -158.2955322265625,
      "logps/rejected": -466.1955871582031,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.9580343711206163e-07,
      "logits/chosen": -2.6071786880493164,
      "logits/rejected": -2.561267137527466,
      "logps/chosen": -257.0948791503906,
      "logps/pi_response": -139.04913330078125,
      "logps/ref_response": -134.45907592773438,
      "logps/rejected": -405.8514709472656,
      "loss": 0.674,
      "rewards/accuracies": 0.5798611044883728,
      "rewards/chosen": -0.08496704697608948,
      "rewards/margins": 0.058161403983831406,
      "rewards/rejected": -0.14312845468521118,
      "step": 10
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.512474502277316e-07,
      "logits/chosen": -2.5841031074523926,
      "logits/rejected": -2.4879150390625,
      "logps/chosen": -301.86749267578125,
      "logps/pi_response": -155.46690368652344,
      "logps/ref_response": -146.185791015625,
      "logps/rejected": -503.9505920410156,
      "loss": 0.5738,
      "rewards/accuracies": 0.746874988079071,
      "rewards/chosen": -0.4705522954463959,
      "rewards/margins": 0.5787664651870728,
      "rewards/rejected": -1.049318790435791,
      "step": 20
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.721469846003722e-07,
      "logits/chosen": -2.410111904144287,
      "logits/rejected": -2.3357787132263184,
      "logps/chosen": -362.79449462890625,
      "logps/pi_response": -168.09576416015625,
      "logps/ref_response": -146.45614624023438,
      "logps/rejected": -546.7628784179688,
      "loss": 0.52,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": -0.852766215801239,
      "rewards/margins": 0.7539817094802856,
      "rewards/rejected": -1.6067478656768799,
      "step": 30
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.549021965852197e-08,
      "logits/chosen": -2.417564630508423,
      "logits/rejected": -2.3170862197875977,
      "logps/chosen": -331.92822265625,
      "logps/pi_response": -158.364013671875,
      "logps/ref_response": -145.10653686523438,
      "logps/rejected": -582.767822265625,
      "loss": 0.5142,
      "rewards/accuracies": 0.746874988079071,
      "rewards/chosen": -0.6605563759803772,
      "rewards/margins": 0.9509684443473816,
      "rewards/rejected": -1.6115245819091797,
      "step": 40
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.0843458234896666e-08,
      "logits/chosen": -2.4105136394500732,
      "logits/rejected": -2.3534531593322754,
      "logps/chosen": -345.56024169921875,
      "logps/pi_response": -149.22549438476562,
      "logps/ref_response": -138.38841247558594,
      "logps/rejected": -550.0933227539062,
      "loss": 0.5322,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": -0.6944135427474976,
      "rewards/margins": 0.7579292058944702,
      "rewards/rejected": -1.4523428678512573,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5597076092736196,
      "train_runtime": 3500.8727,
      "train_samples_per_second": 4.365,
      "train_steps_per_second": 0.017
    }
  ],
  "logging_steps": 10,
  "max_steps": 59,
  "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
}