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": 0.016736401673640166,
      "grad_norm": 90.20505695600441,
      "learning_rate": 1.6666666666666664e-08,
      "logits/chosen": -2.38896107673645,
      "logits/rejected": -2.3177852630615234,
      "logps/chosen": -330.9396057128906,
      "logps/pi_response": -201.37353515625,
      "logps/ref_response": -201.37353515625,
      "logps/rejected": -629.0294189453125,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.16736401673640167,
      "grad_norm": 88.05892874097484,
      "learning_rate": 9.860114570402053e-08,
      "logits/chosen": -2.368415355682373,
      "logits/rejected": -2.2932443618774414,
      "logps/chosen": -344.291015625,
      "logps/pi_response": -186.70086669921875,
      "logps/ref_response": -186.3529510498047,
      "logps/rejected": -626.49609375,
      "loss": 0.6841,
      "rewards/accuracies": 0.5520833134651184,
      "rewards/chosen": -0.014421042054891586,
      "rewards/margins": 0.019680418074131012,
      "rewards/rejected": -0.0341014601290226,
      "step": 10
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 51.86986237888539,
      "learning_rate": 8.374915007591053e-08,
      "logits/chosen": -2.3157577514648438,
      "logits/rejected": -2.2570173740386963,
      "logps/chosen": -373.56463623046875,
      "logps/pi_response": -199.1042938232422,
      "logps/ref_response": -192.49978637695312,
      "logps/rejected": -657.2823486328125,
      "loss": 0.6036,
      "rewards/accuracies": 0.715624988079071,
      "rewards/chosen": -0.22236740589141846,
      "rewards/margins": 0.26518240571022034,
      "rewards/rejected": -0.4875497817993164,
      "step": 20
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 31.24867378826106,
      "learning_rate": 5.738232820012406e-08,
      "logits/chosen": -2.2534313201904297,
      "logits/rejected": -2.2094740867614746,
      "logps/chosen": -364.76763916015625,
      "logps/pi_response": -192.7183837890625,
      "logps/ref_response": -182.37026977539062,
      "logps/rejected": -719.8524169921875,
      "loss": 0.5456,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": -0.36350375413894653,
      "rewards/margins": 0.6372902393341064,
      "rewards/rejected": -1.0007940530776978,
      "step": 30
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 33.556449990736525,
      "learning_rate": 2.8496739886173992e-08,
      "logits/chosen": -2.2471494674682617,
      "logits/rejected": -2.173327684402466,
      "logps/chosen": -401.1842346191406,
      "logps/pi_response": -211.22659301757812,
      "logps/ref_response": -196.47024536132812,
      "logps/rejected": -772.913818359375,
      "loss": 0.5225,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": -0.5726088881492615,
      "rewards/margins": 0.8591602444648743,
      "rewards/rejected": -1.4317691326141357,
      "step": 40
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 74.11847038885749,
      "learning_rate": 6.947819411632222e-09,
      "logits/chosen": -2.226001262664795,
      "logits/rejected": -2.169133186340332,
      "logps/chosen": -440.11309814453125,
      "logps/pi_response": -208.1488494873047,
      "logps/ref_response": -190.44863891601562,
      "logps/rejected": -818.5208740234375,
      "loss": 0.5199,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": -0.7150267362594604,
      "rewards/margins": 0.9438616633415222,
      "rewards/rejected": -1.6588884592056274,
      "step": 50
    },
    {
      "epoch": 0.9874476987447699,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.5673195386337022,
      "train_runtime": 2609.0444,
      "train_samples_per_second": 5.858,
      "train_steps_per_second": 0.023
    }
  ],
  "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
}