trainer_state-gemma-bnb8-QLORA-super_glue-rte-sequence_classification.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.12,
  "eval_steps": 1,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.512,
      "grad_norm": 262.93804931640625,
      "learning_rate": 2.5e-05,
      "loss": 2.8651,
      "step": 1
    },
    {
      "epoch": 0.512,
      "eval_accuracy": 0.5,
      "eval_loss": 2.815425395965576,
      "eval_runtime": 40.3552,
      "eval_samples_per_second": 6.195,
      "eval_steps_per_second": 0.793,
      "step": 1
    },
    {
      "epoch": 1.024,
      "grad_norm": 265.3814697265625,
      "learning_rate": 5e-05,
      "loss": 2.949,
      "step": 2
    },
    {
      "epoch": 1.024,
      "eval_accuracy": 0.516,
      "eval_loss": 1.7428507804870605,
      "eval_runtime": 40.3843,
      "eval_samples_per_second": 6.191,
      "eval_steps_per_second": 0.792,
      "step": 2
    },
    {
      "epoch": 1.536,
      "grad_norm": 178.00296020507812,
      "learning_rate": 4.375e-05,
      "loss": 1.8353,
      "step": 3
    },
    {
      "epoch": 1.536,
      "eval_accuracy": 0.532,
      "eval_loss": 2.163874626159668,
      "eval_runtime": 40.4008,
      "eval_samples_per_second": 6.188,
      "eval_steps_per_second": 0.792,
      "step": 3
    },
    {
      "epoch": 2.048,
      "grad_norm": 195.80059814453125,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 2.1822,
      "step": 4
    },
    {
      "epoch": 2.048,
      "eval_accuracy": 0.52,
      "eval_loss": 2.378251075744629,
      "eval_runtime": 40.452,
      "eval_samples_per_second": 6.18,
      "eval_steps_per_second": 0.791,
      "step": 4
    },
    {
      "epoch": 2.56,
      "grad_norm": 209.93360900878906,
      "learning_rate": 3.125e-05,
      "loss": 2.2654,
      "step": 5
    },
    {
      "epoch": 2.56,
      "eval_accuracy": 0.532,
      "eval_loss": 1.9118725061416626,
      "eval_runtime": 40.2282,
      "eval_samples_per_second": 6.215,
      "eval_steps_per_second": 0.795,
      "step": 5
    },
    {
      "epoch": 3.072,
      "grad_norm": 207.84954833984375,
      "learning_rate": 2.5e-05,
      "loss": 1.9124,
      "step": 6
    },
    {
      "epoch": 3.072,
      "eval_accuracy": 0.58,
      "eval_loss": 1.4134275913238525,
      "eval_runtime": 40.2581,
      "eval_samples_per_second": 6.21,
      "eval_steps_per_second": 0.795,
      "step": 6
    },
    {
      "epoch": 3.584,
      "grad_norm": 128.61007690429688,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 1.2817,
      "step": 7
    },
    {
      "epoch": 3.584,
      "eval_accuracy": 0.636,
      "eval_loss": 1.1336413621902466,
      "eval_runtime": 40.2353,
      "eval_samples_per_second": 6.213,
      "eval_steps_per_second": 0.795,
      "step": 7
    },
    {
      "epoch": 4.096,
      "grad_norm": 67.19721984863281,
      "learning_rate": 1.25e-05,
      "loss": 0.974,
      "step": 8
    },
    {
      "epoch": 4.096,
      "eval_accuracy": 0.636,
      "eval_loss": 1.031956672668457,
      "eval_runtime": 40.2519,
      "eval_samples_per_second": 6.211,
      "eval_steps_per_second": 0.795,
      "step": 8
    },
    {
      "epoch": 4.608,
      "grad_norm": 11.46958065032959,
      "learning_rate": 6.25e-06,
      "loss": 0.809,
      "step": 9
    },
    {
      "epoch": 4.608,
      "eval_accuracy": 0.636,
      "eval_loss": 1.0150794982910156,
      "eval_runtime": 40.2477,
      "eval_samples_per_second": 6.212,
      "eval_steps_per_second": 0.795,
      "step": 9
    },
    {
      "epoch": 5.12,
      "grad_norm": 27.90303611755371,
      "learning_rate": 0.0,
      "loss": 0.8105,
      "step": 10
    },
    {
      "epoch": 5.12,
      "eval_accuracy": 0.64,
      "eval_loss": 1.0157219171524048,
      "eval_runtime": 40.2183,
      "eval_samples_per_second": 6.216,
      "eval_steps_per_second": 0.796,
      "step": 10
    },
    {
      "epoch": 5.12,
      "step": 10,
      "total_flos": 3.5290299719122944e+16,
      "train_loss": 1.7884618103504182,
      "train_runtime": 3021.7281,
      "train_samples_per_second": 3.309,
      "train_steps_per_second": 0.003
    }
  ],
  "logging_steps": 1,
  "max_steps": 10,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 3.5290299719122944e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}