checkpoint-5016/trainer_state.json

{
  "best_metric": 0.9192785620689392,
  "best_model_checkpoint": "facial_emotions_image_detection_rafdb_microsoft_vit/checkpoint-5016",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 5016,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.8,
      "grad_norm": 14.677521705627441,
      "learning_rate": 2.7281514297221103e-06,
      "loss": 1.7706,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.45075587486903157,
      "eval_loss": 1.4894078969955444,
      "eval_runtime": 69.9167,
      "eval_samples_per_second": 191.113,
      "eval_steps_per_second": 11.957,
      "step": 627
    },
    {
      "epoch": 1.59,
      "grad_norm": 17.35078239440918,
      "learning_rate": 2.4260974627466776e-06,
      "loss": 1.4285,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5424337674000899,
      "eval_loss": 1.2471543550491333,
      "eval_runtime": 65.1157,
      "eval_samples_per_second": 205.204,
      "eval_steps_per_second": 12.839,
      "step": 1254
    },
    {
      "epoch": 2.39,
      "grad_norm": 15.841980934143066,
      "learning_rate": 2.1240434957712445e-06,
      "loss": 1.2563,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5873372249663225,
      "eval_loss": 1.1151762008666992,
      "eval_runtime": 65.1132,
      "eval_samples_per_second": 205.212,
      "eval_steps_per_second": 12.839,
      "step": 1881
    },
    {
      "epoch": 3.19,
      "grad_norm": 24.84258270263672,
      "learning_rate": 1.8219895287958114e-06,
      "loss": 1.1575,
      "step": 2000
    },
    {
      "epoch": 3.99,
      "grad_norm": 14.571057319641113,
      "learning_rate": 1.5199355618203787e-06,
      "loss": 1.0741,
      "step": 2500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6200419098937285,
      "eval_loss": 1.0384080410003662,
      "eval_runtime": 65.4313,
      "eval_samples_per_second": 204.214,
      "eval_steps_per_second": 12.777,
      "step": 2508
    },
    {
      "epoch": 4.78,
      "grad_norm": 19.834932327270508,
      "learning_rate": 1.2178815948449456e-06,
      "loss": 1.0238,
      "step": 3000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6430923514443946,
      "eval_loss": 0.981325626373291,
      "eval_runtime": 66.0473,
      "eval_samples_per_second": 202.31,
      "eval_steps_per_second": 12.658,
      "step": 3135
    },
    {
      "epoch": 5.58,
      "grad_norm": 17.73134994506836,
      "learning_rate": 9.158276278695127e-07,
      "loss": 0.9837,
      "step": 3500
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6525969166292471,
      "eval_loss": 0.9464946985244751,
      "eval_runtime": 66.383,
      "eval_samples_per_second": 201.286,
      "eval_steps_per_second": 12.594,
      "step": 3762
    },
    {
      "epoch": 6.38,
      "grad_norm": 14.271998405456543,
      "learning_rate": 6.137736608940798e-07,
      "loss": 0.9694,
      "step": 4000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6600059871276756,
      "eval_loss": 0.9257609248161316,
      "eval_runtime": 66.6168,
      "eval_samples_per_second": 200.58,
      "eval_steps_per_second": 12.549,
      "step": 4389
    },
    {
      "epoch": 7.18,
      "grad_norm": 26.607189178466797,
      "learning_rate": 3.117196939186468e-07,
      "loss": 0.9391,
      "step": 4500
    },
    {
      "epoch": 7.97,
      "grad_norm": 20.43568992614746,
      "learning_rate": 9.665726943213855e-09,
      "loss": 0.9235,
      "step": 5000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.663897620116749,
      "eval_loss": 0.9192785620689392,
      "eval_runtime": 64.6255,
      "eval_samples_per_second": 206.761,
      "eval_steps_per_second": 12.936,
      "step": 5016
    }
  ],
  "logging_steps": 500,
  "max_steps": 5016,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 500,
  "total_flos": 3.98586664581061e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}