trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "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": 24.253913625492473,
      "learning_rate": 1e-08,
      "logits/chosen": -2.3481905460357666,
      "logits/rejected": -2.0932602882385254,
      "logps/chosen": -244.2006378173828,
      "logps/rejected": -278.7900390625,
      "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": 26.023179273047976,
      "learning_rate": 1e-07,
      "logits/chosen": -2.2961385250091553,
      "logits/rejected": -2.423363208770752,
      "logps/chosen": -266.5706481933594,
      "logps/rejected": -210.25784301757812,
      "loss": 0.6931,
      "rewards/accuracies": 0.4513888955116272,
      "rewards/chosen": 4.0644288674229756e-05,
      "rewards/margins": 0.00030885564046911895,
      "rewards/rejected": -0.000268211355432868,
      "step": 10
    },
    {
      "epoch": 0.21,
      "grad_norm": 23.705405318866816,
      "learning_rate": 9.65436874322102e-08,
      "logits/chosen": -2.417802095413208,
      "logits/rejected": -2.517642021179199,
      "logps/chosen": -241.8213653564453,
      "logps/rejected": -222.079345703125,
      "loss": 0.6916,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": 0.003412181045860052,
      "rewards/margins": 0.0023693442344665527,
      "rewards/rejected": 0.001042836345732212,
      "step": 20
    },
    {
      "epoch": 0.32,
      "grad_norm": 24.678482409057676,
      "learning_rate": 8.665259359149131e-08,
      "logits/chosen": -2.2216708660125732,
      "logits/rejected": -2.7527403831481934,
      "logps/chosen": -277.8819885253906,
      "logps/rejected": -184.9886016845703,
      "loss": 0.6877,
      "rewards/accuracies": 0.7437499761581421,
      "rewards/chosen": 0.012889519333839417,
      "rewards/margins": 0.011960303410887718,
      "rewards/rejected": 0.0009292167378589511,
      "step": 30
    },
    {
      "epoch": 0.43,
      "grad_norm": 23.771630985115603,
      "learning_rate": 7.16941869558779e-08,
      "logits/chosen": -2.4029767513275146,
      "logits/rejected": -2.3705716133117676,
      "logps/chosen": -188.0555419921875,
      "logps/rejected": -197.7487030029297,
      "loss": 0.6846,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": 0.016015395522117615,
      "rewards/margins": 0.01416720449924469,
      "rewards/rejected": 0.0018481904407963157,
      "step": 40
    },
    {
      "epoch": 0.53,
      "grad_norm": 25.104287071663315,
      "learning_rate": 5.373650467932121e-08,
      "logits/chosen": -2.374539613723755,
      "logits/rejected": -2.6364951133728027,
      "logps/chosen": -218.8348388671875,
      "logps/rejected": -195.1355743408203,
      "loss": 0.6801,
      "rewards/accuracies": 0.824999988079071,
      "rewards/chosen": 0.02972819283604622,
      "rewards/margins": 0.025985542684793472,
      "rewards/rejected": 0.0037426550406962633,
      "step": 50
    },
    {
      "epoch": 0.64,
      "grad_norm": 22.342352206662035,
      "learning_rate": 3.5262241279454787e-08,
      "logits/chosen": -2.220731735229492,
      "logits/rejected": -2.57670259475708,
      "logps/chosen": -229.725341796875,
      "logps/rejected": -189.80429077148438,
      "loss": 0.6765,
      "rewards/accuracies": 0.8812500238418579,
      "rewards/chosen": 0.03678583353757858,
      "rewards/margins": 0.03224078193306923,
      "rewards/rejected": 0.004545052535831928,
      "step": 60
    },
    {
      "epoch": 0.74,
      "grad_norm": 26.142147699838866,
      "learning_rate": 1.8825509907063325e-08,
      "logits/chosen": -2.286602020263672,
      "logits/rejected": -2.6010844707489014,
      "logps/chosen": -233.9541778564453,
      "logps/rejected": -184.05123901367188,
      "loss": 0.6729,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": 0.04538162797689438,
      "rewards/margins": 0.04122646898031235,
      "rewards/rejected": 0.004155161790549755,
      "step": 70
    },
    {
      "epoch": 0.85,
      "grad_norm": 24.562196541892764,
      "learning_rate": 6.698729810778064e-09,
      "logits/chosen": -2.2746427059173584,
      "logits/rejected": -2.5367631912231445,
      "logps/chosen": -236.80679321289062,
      "logps/rejected": -204.65162658691406,
      "loss": 0.6711,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": 0.050890929996967316,
      "rewards/margins": 0.04626064375042915,
      "rewards/rejected": 0.004630280192941427,
      "step": 80
    },
    {
      "epoch": 0.96,
      "grad_norm": 24.603581055332214,
      "learning_rate": 5.584586887435739e-10,
      "logits/chosen": -2.228706121444702,
      "logits/rejected": -2.7618346214294434,
      "logps/chosen": -248.544921875,
      "logps/rejected": -183.44293212890625,
      "loss": 0.6724,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.052454717457294464,
      "rewards/margins": 0.04798055440187454,
      "rewards/rejected": 0.004474165849387646,
      "step": 90
    },
    {
      "epoch": 1.0,
      "step": 94,
      "total_flos": 0.0,
      "train_loss": 0.680875283606509,
      "train_runtime": 1044.3353,
      "train_samples_per_second": 5.758,
      "train_steps_per_second": 0.09
    }
  ],
  "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
}