trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 80,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-08,
      "logits/chosen": -2.8984241485595703,
      "logits/rejected": -2.9192605018615723,
      "logps/chosen": -223.2982635498047,
      "logps/pi_response": -113.8905029296875,
      "logps/ref_response": -113.8905029296875,
      "logps/rejected": -200.18243408203125,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.990486745229364e-07,
      "logits/chosen": -2.762209892272949,
      "logits/rejected": -2.7415664196014404,
      "logps/chosen": -217.4150390625,
      "logps/pi_response": -115.45431518554688,
      "logps/ref_response": -115.2744140625,
      "logps/rejected": -178.28346252441406,
      "loss": 0.6927,
      "rewards/accuracies": 0.5069444179534912,
      "rewards/chosen": 0.0064659034833312035,
      "rewards/margins": 0.004789570812135935,
      "rewards/rejected": 0.0016763320891186595,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.6650635094610966e-07,
      "logits/chosen": -2.781243085861206,
      "logits/rejected": -2.7348623275756836,
      "logps/chosen": -220.1856231689453,
      "logps/pi_response": -134.96209716796875,
      "logps/ref_response": -125.4894027709961,
      "logps/rejected": -214.1865234375,
      "loss": 0.6816,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": -0.04695942997932434,
      "rewards/margins": 0.025210771709680557,
      "rewards/rejected": -0.0721702054142952,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.933941090877615e-07,
      "logits/chosen": -2.619565486907959,
      "logits/rejected": -2.608074903488159,
      "logps/chosen": -220.36355590820312,
      "logps/pi_response": -125.2795181274414,
      "logps/ref_response": -107.89570617675781,
      "logps/rejected": -210.0551300048828,
      "loss": 0.6602,
      "rewards/accuracies": 0.6187499761581421,
      "rewards/chosen": -0.08404215425252914,
      "rewards/margins": 0.10765181481838226,
      "rewards/rejected": -0.1916939914226532,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.934120444167326e-07,
      "logits/chosen": -2.649507522583008,
      "logits/rejected": -2.6370885372161865,
      "logps/chosen": -236.1080780029297,
      "logps/pi_response": -141.00830078125,
      "logps/ref_response": -108.48822021484375,
      "logps/rejected": -220.882080078125,
      "loss": 0.6488,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -0.2109258472919464,
      "rewards/margins": 0.1395840346813202,
      "rewards/rejected": -0.3505098819732666,
      "step": 40
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8529523872436977e-07,
      "logits/chosen": -2.5717177391052246,
      "logits/rejected": -2.5684621334075928,
      "logps/chosen": -217.80819702148438,
      "logps/pi_response": -137.82199096679688,
      "logps/ref_response": -103.4356918334961,
      "logps/rejected": -215.01364135742188,
      "loss": 0.6332,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": -0.18964743614196777,
      "rewards/margins": 0.15112996101379395,
      "rewards/rejected": -0.3407774269580841,
      "step": 50
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.930309757836516e-08,
      "logits/chosen": -2.5936081409454346,
      "logits/rejected": -2.5533876419067383,
      "logps/chosen": -255.20663452148438,
      "logps/pi_response": -153.9727325439453,
      "logps/ref_response": -109.31465148925781,
      "logps/rejected": -229.056396484375,
      "loss": 0.632,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.3040311932563782,
      "rewards/margins": 0.177860826253891,
      "rewards/rejected": -0.48189201951026917,
      "step": 60
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.3423053240837514e-08,
      "logits/chosen": -2.5863394737243652,
      "logits/rejected": -2.5548641681671143,
      "logps/chosen": -238.87612915039062,
      "logps/pi_response": -155.53675842285156,
      "logps/ref_response": -100.11727142333984,
      "logps/rejected": -238.0329132080078,
      "loss": 0.6206,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.3792975842952728,
      "rewards/margins": 0.22778081893920898,
      "rewards/rejected": -0.6070784330368042,
      "step": 70
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "logits/chosen": -2.6336300373077393,
      "logits/rejected": -2.6055710315704346,
      "logps/chosen": -246.85049438476562,
      "logps/pi_response": -163.44818115234375,
      "logps/ref_response": -113.96580505371094,
      "logps/rejected": -236.2712860107422,
      "loss": 0.6235,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -0.3604409098625183,
      "rewards/margins": 0.16606533527374268,
      "rewards/rejected": -0.526506245136261,
      "step": 80
    },
    {
      "epoch": 1.0,
      "step": 80,
      "total_flos": 0.0,
      "train_loss": 0.6490777730941772,
      "train_runtime": 2263.5288,
      "train_samples_per_second": 4.501,
      "train_steps_per_second": 0.035
    }
  ],
  "logging_steps": 10,
  "max_steps": 80,
  "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
}