{
  "best_metric": 0.3777986466884613,
  "best_model_checkpoint": "./zephyr/09-04-24-Weni-WeniGPT-Agents-Zephyr-1.0.13-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": 2.111046075820923,
      "learning_rate": 0.00018,
      "loss": 0.4439,
      "step": 20,
      "train/kl": 8.516629219055176,
      "train/logps/chosen": -296.32746550324674,
      "train/logps/rejected": -277.9095679593373,
      "train/rewards/chosen": 0.7222159744857194,
      "train/rewards/margins": 0.9555325568552365,
      "train/rewards/rejected": -0.23331658236951713
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2633113861083984,
      "learning_rate": 0.00015142857142857143,
      "loss": 0.4194,
      "step": 40,
      "train/kl": 1.0890424251556396,
      "train/logps/chosen": -307.1073526152482,
      "train/logps/rejected": -314.9674406424581,
      "train/rewards/chosen": -1.5282932579094637,
      "train/rewards/margins": 1.4390720282942722,
      "train/rewards/rejected": -2.967365286203736
    },
    {
      "epoch": 0.34,
      "eval/kl": 15.062309265136719,
      "eval/logps/chosen": -259.4809639084507,
      "eval/logps/rejected": -256.0586679193038,
      "eval/rewards/chosen": 2.41808867118728,
      "eval/rewards/margins": 1.5870417096564255,
      "eval/rewards/rejected": 0.8310469615308544,
      "eval_loss": 0.42071497440338135,
      "eval_runtime": 140.3611,
      "eval_samples_per_second": 2.137,
      "eval_steps_per_second": 0.534,
      "step": 50
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.257838487625122,
      "learning_rate": 0.00012285714285714287,
      "loss": 0.398,
      "step": 60,
      "train/kl": 14.152316093444824,
      "train/logps/chosen": -258.6226310483871,
      "train/logps/rejected": -281.4683475378788,
      "train/rewards/chosen": 2.234444304435484,
      "train/rewards/margins": 1.6334723195722025,
      "train/rewards/rejected": 0.6009719848632813
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.802126884460449,
      "learning_rate": 9.428571428571429e-05,
      "loss": 0.381,
      "step": 80,
      "train/kl": 3.5482826232910156,
      "train/logps/chosen": -274.61489125844594,
      "train/logps/rejected": -313.1891124636628,
      "train/rewards/chosen": 0.9537969022183805,
      "train/rewards/margins": 3.017533396415782,
      "train/rewards/rejected": -2.0637364941974017
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.998615026473999,
      "learning_rate": 6.571428571428571e-05,
      "loss": 0.3603,
      "step": 100,
      "train/kl": 6.8296308517456055,
      "train/logps/chosen": -256.6186432453416,
      "train/logps/rejected": -291.86861242138366,
      "train/rewards/chosen": 2.3773075838266693,
      "train/rewards/margins": 3.0153399332272195,
      "train/rewards/rejected": -0.6380323494005503
    },
    {
      "epoch": 0.68,
      "eval/kl": 10.186383247375488,
      "eval/logps/chosen": -251.80529269366198,
      "eval/logps/rejected": -260.74075356012656,
      "eval/rewards/chosen": 3.1856545461735255,
      "eval/rewards/margins": 2.8228144007244302,
      "eval/rewards/rejected": 0.36284014544909515,
      "eval_loss": 0.3777986466884613,
      "eval_runtime": 140.492,
      "eval_samples_per_second": 2.135,
      "eval_steps_per_second": 0.534,
      "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
}