{
  "best_metric": 0.35275527834892273,
  "best_model_checkpoint": "./zephyr/09-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.18-KTO_Hyperparameter search, altering lora params for KTO task.-2_max_steps-145_batch_16_2024-04-09_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": 6.346695899963379,
      "learning_rate": 0.00018142857142857142,
      "loss": 0.467,
      "step": 20,
      "train/kl": 6.117425918579102,
      "train/logps/chosen": -259.1842447916667,
      "train/logps/rejected": -290.1948988970588,
      "train/rewards/chosen": 0.930299072265625,
      "train/rewards/margins": 0.5634524266860065,
      "train/rewards/rejected": 0.36684664557961855
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.604153633117676,
      "learning_rate": 0.00015285714285714287,
      "loss": 0.4224,
      "step": 40,
      "train/kl": 6.080809116363525,
      "train/logps/chosen": -274.32459677419354,
      "train/logps/rejected": -291.6558948863636,
      "train/rewards/chosen": 0.9791939027847782,
      "train/rewards/margins": 1.6478286295692244,
      "train/rewards/rejected": -0.668634726784446
    },
    {
      "epoch": 0.34,
      "eval/kl": 7.511639595031738,
      "eval/logps/chosen": -263.8732394366197,
      "eval/logps/rejected": -268.6064082278481,
      "eval/rewards/chosen": 1.9797810299295775,
      "eval/rewards/margins": 2.403955568071142,
      "eval/rewards/rejected": -0.42417453814156447,
      "eval_loss": 0.3916032016277313,
      "eval_runtime": 141.678,
      "eval_samples_per_second": 2.117,
      "eval_steps_per_second": 0.529,
      "step": 50
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.5113444328308105,
      "learning_rate": 0.00012428571428571428,
      "loss": 0.3832,
      "step": 60,
      "train/kl": 16.34114646911621,
      "train/logps/chosen": -240.63917267628204,
      "train/logps/rejected": -277.463486089939,
      "train/rewards/chosen": 3.0404166197165465,
      "train/rewards/margins": 2.3469540618075815,
      "train/rewards/rejected": 0.6934625579089653
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7619431018829346,
      "learning_rate": 9.571428571428573e-05,
      "loss": 0.3233,
      "step": 80,
      "train/kl": 0.7871202826499939,
      "train/logps/chosen": -294.5624213506711,
      "train/logps/rejected": -336.1468612938597,
      "train/rewards/chosen": 0.5479572987396445,
      "train/rewards/margins": 5.502926202933724,
      "train/rewards/rejected": -4.954968904194079
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.996425151824951,
      "learning_rate": 6.714285714285714e-05,
      "loss": 0.3749,
      "step": 100,
      "train/kl": 6.590612888336182,
      "train/logps/chosen": -265.2372325922819,
      "train/logps/rejected": -300.5130665204678,
      "train/rewards/chosen": 2.635832920970533,
      "train/rewards/margins": 3.7929440163766914,
      "train/rewards/rejected": -1.1571110954061585
    },
    {
      "epoch": 0.68,
      "eval/kl": 2.193509817123413,
      "eval/logps/chosen": -268.1729478433099,
      "eval/logps/rejected": -291.57960838607596,
      "eval/rewards/chosen": 1.5498130690883583,
      "eval/rewards/margins": 4.271308299075008,
      "eval/rewards/rejected": -2.7214952299866497,
      "eval_loss": 0.35275527834892273,
      "eval_runtime": 141.6617,
      "eval_samples_per_second": 2.118,
      "eval_steps_per_second": 0.529,
      "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
}