checkpoint-100/trainer_state.json

{
  "best_metric": 0.47333332896232605,
  "best_model_checkpoint": "./zephyr/05-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.1-KTO_testing kto dataset during training-3_max_steps-145_batch_16_2024-04-05_ppid_9/checkpoint-100",
  "epoch": 0.684931506849315,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.14,
      "grad_norm": 32.5521125793457,
      "kl": 6.568634033203125,
      "learning_rate": 0.00018285714285714286,
      "logps/chosen": -309.11578369140625,
      "logps/rejected": -330.8851013183594,
      "loss": 0.4463,
      "rewards/chosen": -1.5149470567703247,
      "rewards/margins": 0.730856716632843,
      "rewards/rejected": -2.1397347450256348,
      "step": 20
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.194253444671631,
      "kl": 9.558149337768555,
      "learning_rate": 0.00015571428571428572,
      "logps/chosen": -317.7120666503906,
      "logps/rejected": -336.9992980957031,
      "loss": 0.4204,
      "rewards/chosen": -0.900913655757904,
      "rewards/margins": 1.870524287223816,
      "rewards/rejected": -2.7778050899505615,
      "step": 40
    },
    {
      "epoch": 0.34,
      "eval_kl": 0.0,
      "eval_logps/chosen": -471.27984619140625,
      "eval_logps/rejected": -470.8447570800781,
      "eval_loss": 0.47356343269348145,
      "eval_rewards/chosen": -17.10463523864746,
      "eval_rewards/margins": 1.9298614263534546,
      "eval_rewards/rejected": -19.51471710205078,
      "eval_runtime": 137.637,
      "eval_samples_per_second": 2.18,
      "eval_steps_per_second": 0.545,
      "step": 50
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.0,
      "kl": 0.2518434524536133,
      "learning_rate": 0.00012714285714285714,
      "logps/chosen": -638.1253051757812,
      "logps/rejected": -641.34716796875,
      "loss": 0.402,
      "rewards/chosen": -33.93111038208008,
      "rewards/margins": 1.04580819606781,
      "rewards/rejected": -34.0406379699707,
      "step": 60
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 9.857142857142858e-05,
      "logps/chosen": -1467.325927734375,
      "logps/rejected": -1537.5704345703125,
      "loss": 0.45,
      "rewards/chosen": -117.26831817626953,
      "rewards/margins": 9.93100357055664,
      "rewards/rejected": -124.3635025024414,
      "step": 80
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 7e-05,
      "logps/chosen": -2025.484375,
      "logps/rejected": -1882.913818359375,
      "loss": 0.4813,
      "rewards/chosen": -169.48184204101562,
      "rewards/margins": -10.249415397644043,
      "rewards/rejected": -157.5347900390625,
      "step": 100
    },
    {
      "epoch": 0.68,
      "eval_kl": 0.0,
      "eval_logps/chosen": -1703.4486083984375,
      "eval_logps/rejected": -1636.100341796875,
      "eval_loss": 0.47333332896232605,
      "eval_rewards/chosen": -140.32151794433594,
      "eval_rewards/margins": -6.890669822692871,
      "eval_rewards/rejected": -136.040283203125,
      "eval_runtime": 137.631,
      "eval_samples_per_second": 2.18,
      "eval_steps_per_second": 0.545,
      "step": 100
    }
  ],
  "logging_steps": 20,
  "max_steps": 145,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}