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": 7.772396914178464,
      "learning_rate": 8.333333333333333e-08,
      "logits/chosen": -2.807276487350464,
      "logits/rejected": -2.7759768962860107,
      "logps/chosen": -315.42626953125,
      "logps/rejected": -227.5915985107422,
      "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": 7.458310251605103,
      "learning_rate": 4.930057285201027e-07,
      "logits/chosen": -2.755532741546631,
      "logits/rejected": -2.7461793422698975,
      "logps/chosen": -271.9573974609375,
      "logps/rejected": -260.7023620605469,
      "loss": 0.6915,
      "rewards/accuracies": 0.5243055820465088,
      "rewards/chosen": 0.0051252637058496475,
      "rewards/margins": 0.003539684694260359,
      "rewards/rejected": 0.001585578778758645,
      "step": 10
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 6.720707588648352,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -2.786818027496338,
      "logits/rejected": -2.7679667472839355,
      "logps/chosen": -261.4557189941406,
      "logps/rejected": -250.8779754638672,
      "loss": 0.6745,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.03225986286997795,
      "rewards/margins": 0.040614787489175797,
      "rewards/rejected": -0.008354922756552696,
      "step": 20
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 7.434878606336098,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -2.7875802516937256,
      "logits/rejected": -2.7685418128967285,
      "logps/chosen": -293.7994079589844,
      "logps/rejected": -254.1815948486328,
      "loss": 0.6487,
      "rewards/accuracies": 0.6781250238418579,
      "rewards/chosen": -0.02343321219086647,
      "rewards/margins": 0.12680219113826752,
      "rewards/rejected": -0.1502353996038437,
      "step": 30
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 8.154579364951834,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": -2.760192394256592,
      "logits/rejected": -2.737757444381714,
      "logps/chosen": -264.1371154785156,
      "logps/rejected": -255.595458984375,
      "loss": 0.6354,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.11217646300792694,
      "rewards/margins": 0.16067178547382355,
      "rewards/rejected": -0.2728482186794281,
      "step": 40
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 8.586570040476488,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": -2.77009654045105,
      "logits/rejected": -2.7502076625823975,
      "logps/chosen": -279.09228515625,
      "logps/rejected": -287.27227783203125,
      "loss": 0.6202,
      "rewards/accuracies": 0.621874988079071,
      "rewards/chosen": -0.22083595395088196,
      "rewards/margins": 0.1367621123790741,
      "rewards/rejected": -0.35759806632995605,
      "step": 50
    },
    {
      "epoch": 0.9874476987447699,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.6490584389638092,
      "train_runtime": 1634.5367,
      "train_samples_per_second": 9.35,
      "train_steps_per_second": 0.036
    }
  ],
  "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
}