{
  "best_metric": 0.3619329333305359,
  "best_model_checkpoint": "./zephyr/08-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.17-KTO_Hyperparameter search, altering lora params for KTO task.-2_max_steps-145_batch_16_2024-04-08_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": 6.565266132354736,
      "learning_rate": 0.00018142857142857142,
      "loss": 0.3847,
      "step": 20,
      "train/kl": 3.3989148139953613,
      "train/logps/chosen": -298.30405479753523,
      "train/logps/rejected": -317.74027826544943,
      "train/rewards/chosen": 0.4325446276597574,
      "train/rewards/margins": 3.3660581404318783,
      "train/rewards/rejected": -2.933513512772121
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.64931583404541,
      "learning_rate": 0.00015285714285714287,
      "loss": 0.4039,
      "step": 40,
      "train/kl": 0.9900484085083008,
      "train/logps/chosen": -305.07652368012424,
      "train/logps/rejected": -347.4229805424528,
      "train/rewards/chosen": -2.13509221995099,
      "train/rewards/margins": 5.288080256955457,
      "train/rewards/rejected": -7.423172476906447
    },
    {
      "epoch": 0.34,
      "eval/kl": 0.7344650030136108,
      "eval/logps/chosen": -329.9394806338028,
      "eval/logps/rejected": -444.9535700158228,
      "eval/rewards/chosen": -4.62774013465559,
      "eval/rewards/margins": 13.430626471851532,
      "eval/rewards/rejected": -18.05836660650712,
      "eval_loss": 0.37672173976898193,
      "eval_runtime": 140.2349,
      "eval_samples_per_second": 2.139,
      "eval_steps_per_second": 0.535,
      "step": 50
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.288690567016602,
      "learning_rate": 0.00012714285714285714,
      "loss": 0.3602,
      "step": 60,
      "train/kl": 2.266563653945923,
      "train/logps/chosen": -318.9802876655629,
      "train/logps/rejected": -473.3688517011834,
      "train/rewards/chosen": -3.7534725366049253,
      "train/rewards/margins": 16.50871138073975,
      "train/rewards/rejected": -20.262183917344675
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.295706748962402,
      "learning_rate": 9.857142857142858e-05,
      "loss": 0.3496,
      "step": 80,
      "train/kl": 0.8752914667129517,
      "train/logps/chosen": -335.9194670376712,
      "train/logps/rejected": -405.97503591954023,
      "train/rewards/chosen": -3.1786450947800726,
      "train/rewards/margins": 7.811195007141622,
      "train/rewards/rejected": -10.989840101921695
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.850632667541504,
      "learning_rate": 7e-05,
      "loss": 0.302,
      "step": 100,
      "train/kl": 0.4592212736606598,
      "train/logps/chosen": -277.03286903782896,
      "train/logps/rejected": -392.9672154017857,
      "train/rewards/chosen": 0.3360620799817537,
      "train/rewards/margins": 9.165781260134283,
      "train/rewards/rejected": -8.82971918015253
    },
    {
      "epoch": 0.68,
      "eval/kl": 3.9467480182647705,
      "eval/logps/chosen": -265.8084286971831,
      "eval/logps/rejected": -297.73909711234177,
      "eval/rewards/chosen": 1.7853647151463468,
      "eval/rewards/margins": 5.122287276212407,
      "eval/rewards/rejected": -3.3369225610660602,
      "eval_loss": 0.3619329333305359,
      "eval_runtime": 140.261,
      "eval_samples_per_second": 2.139,
      "eval_steps_per_second": 0.535,
      "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
}