trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9947089947089947,
  "eval_steps": 500,
  "global_step": 94,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 25.005165779900267,
      "learning_rate": 1e-08,
      "logits/chosen": -1.9501205682754517,
      "logits/rejected": -2.513594388961792,
      "logps/chosen": -348.5884704589844,
      "logps/rejected": -166.58517456054688,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.68906488321,
      "learning_rate": 1e-07,
      "logits/chosen": -2.640984535217285,
      "logits/rejected": -2.219906806945801,
      "logps/chosen": -213.95584106445312,
      "logps/rejected": -198.17874145507812,
      "loss": 0.6932,
      "rewards/accuracies": 0.4930555522441864,
      "rewards/chosen": 0.0007135343039408326,
      "rewards/margins": 0.0006534373969770968,
      "rewards/rejected": 6.009703065501526e-05,
      "step": 10
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.52231020802604,
      "learning_rate": 9.65436874322102e-08,
      "logits/chosen": -2.397062301635742,
      "logits/rejected": -2.3303606510162354,
      "logps/chosen": -257.8389587402344,
      "logps/rejected": -222.90444946289062,
      "loss": 0.6917,
      "rewards/accuracies": 0.6187499761581421,
      "rewards/chosen": 0.0030278589110821486,
      "rewards/margins": 0.002860091160982847,
      "rewards/rejected": 0.00016776802658569068,
      "step": 20
    },
    {
      "epoch": 0.32,
      "grad_norm": 24.931881082321315,
      "learning_rate": 8.665259359149131e-08,
      "logits/chosen": -2.534593105316162,
      "logits/rejected": -2.4346184730529785,
      "logps/chosen": -227.6776885986328,
      "logps/rejected": -204.8966064453125,
      "loss": 0.689,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.00954088568687439,
      "rewards/margins": 0.00823338981717825,
      "rewards/rejected": 0.0013074951712042093,
      "step": 30
    },
    {
      "epoch": 0.42,
      "grad_norm": 24.657287539381663,
      "learning_rate": 7.16941869558779e-08,
      "logits/chosen": -2.358189105987549,
      "logits/rejected": -2.440410614013672,
      "logps/chosen": -230.328857421875,
      "logps/rejected": -210.7056427001953,
      "loss": 0.6846,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": 0.018901044502854347,
      "rewards/margins": 0.01690484955906868,
      "rewards/rejected": 0.0019961954094469547,
      "step": 40
    },
    {
      "epoch": 0.53,
      "grad_norm": 24.026917311824274,
      "learning_rate": 5.373650467932121e-08,
      "logits/chosen": -2.379296064376831,
      "logits/rejected": -2.6683709621429443,
      "logps/chosen": -233.249267578125,
      "logps/rejected": -199.204833984375,
      "loss": 0.6805,
      "rewards/accuracies": 0.8187500238418579,
      "rewards/chosen": 0.02693324163556099,
      "rewards/margins": 0.024523768573999405,
      "rewards/rejected": 0.002409472828730941,
      "step": 50
    },
    {
      "epoch": 0.63,
      "grad_norm": 23.069042390909715,
      "learning_rate": 3.5262241279454787e-08,
      "logits/chosen": -2.233121156692505,
      "logits/rejected": -2.644624710083008,
      "logps/chosen": -250.5662078857422,
      "logps/rejected": -173.79067993164062,
      "loss": 0.6762,
      "rewards/accuracies": 0.90625,
      "rewards/chosen": 0.03840692713856697,
      "rewards/margins": 0.03592415899038315,
      "rewards/rejected": 0.0024827648885548115,
      "step": 60
    },
    {
      "epoch": 0.74,
      "grad_norm": 23.60279896856753,
      "learning_rate": 1.8825509907063325e-08,
      "logits/chosen": -2.354218006134033,
      "logits/rejected": -2.420661449432373,
      "logps/chosen": -241.6086883544922,
      "logps/rejected": -207.097900390625,
      "loss": 0.6725,
      "rewards/accuracies": 0.824999988079071,
      "rewards/chosen": 0.04547766596078873,
      "rewards/margins": 0.04097073897719383,
      "rewards/rejected": 0.004506924655288458,
      "step": 70
    },
    {
      "epoch": 0.85,
      "grad_norm": 23.640093404183123,
      "learning_rate": 6.698729810778064e-09,
      "logits/chosen": -2.2659950256347656,
      "logits/rejected": -2.467071056365967,
      "logps/chosen": -243.5118408203125,
      "logps/rejected": -205.4396209716797,
      "loss": 0.6739,
      "rewards/accuracies": 0.8812500238418579,
      "rewards/chosen": 0.0479864627122879,
      "rewards/margins": 0.04288201406598091,
      "rewards/rejected": 0.005104447714984417,
      "step": 80
    },
    {
      "epoch": 0.95,
      "grad_norm": 22.97821730211384,
      "learning_rate": 5.584586887435739e-10,
      "logits/chosen": -2.3436455726623535,
      "logits/rejected": -2.367281436920166,
      "logps/chosen": -226.5314178466797,
      "logps/rejected": -200.30862426757812,
      "loss": 0.6745,
      "rewards/accuracies": 0.8374999761581421,
      "rewards/chosen": 0.043580617755651474,
      "rewards/margins": 0.037381939589977264,
      "rewards/rejected": 0.006198678631335497,
      "step": 90
    },
    {
      "epoch": 0.99,
      "step": 94,
      "total_flos": 0.0,
      "train_loss": 0.6816894855905087,
      "train_runtime": 1070.9433,
      "train_samples_per_second": 5.639,
      "train_steps_per_second": 0.088
    }
  ],
  "logging_steps": 10,
  "max_steps": 94,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}