finetuned-vit-flowers / trainer_state.json
manoh2f2's picture
End of training
6191630
{
"best_metric": 0.972972972972973,
"best_model_checkpoint": "finetuned-vit-flowers/checkpoint-36",
"epoch": 2.958904109589041,
"eval_steps": 500,
"global_step": 108,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.27,
"learning_rate": 4.545454545454546e-05,
"loss": 0.1457,
"step": 10
},
{
"epoch": 0.55,
"learning_rate": 4.536082474226804e-05,
"loss": 0.1297,
"step": 20
},
{
"epoch": 0.82,
"learning_rate": 4.020618556701031e-05,
"loss": 0.1236,
"step": 30
},
{
"epoch": 0.99,
"eval_accuracy": 0.972972972972973,
"eval_loss": 0.1508660614490509,
"eval_runtime": 5.2228,
"eval_samples_per_second": 49.59,
"eval_steps_per_second": 3.255,
"step": 36
},
{
"epoch": 1.1,
"learning_rate": 3.5051546391752576e-05,
"loss": 0.1215,
"step": 40
},
{
"epoch": 1.37,
"learning_rate": 2.9896907216494846e-05,
"loss": 0.1158,
"step": 50
},
{
"epoch": 1.64,
"learning_rate": 2.4742268041237116e-05,
"loss": 0.0942,
"step": 60
},
{
"epoch": 1.92,
"learning_rate": 1.9587628865979382e-05,
"loss": 0.1043,
"step": 70
},
{
"epoch": 2.0,
"eval_accuracy": 0.972972972972973,
"eval_loss": 0.12350893765687943,
"eval_runtime": 5.8451,
"eval_samples_per_second": 44.311,
"eval_steps_per_second": 2.908,
"step": 73
},
{
"epoch": 2.19,
"learning_rate": 1.4432989690721649e-05,
"loss": 0.148,
"step": 80
},
{
"epoch": 2.47,
"learning_rate": 9.278350515463918e-06,
"loss": 0.1162,
"step": 90
},
{
"epoch": 2.74,
"learning_rate": 4.123711340206186e-06,
"loss": 0.1077,
"step": 100
},
{
"epoch": 2.96,
"eval_accuracy": 0.9652509652509652,
"eval_loss": 0.1365247368812561,
"eval_runtime": 5.8827,
"eval_samples_per_second": 44.028,
"eval_steps_per_second": 2.89,
"step": 108
},
{
"epoch": 2.96,
"step": 108,
"total_flos": 5.347091052398592e+17,
"train_loss": 0.12152913654292072,
"train_runtime": 316.4284,
"train_samples_per_second": 22.09,
"train_steps_per_second": 0.341
}
],
"logging_steps": 10,
"max_steps": 108,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 5.347091052398592e+17,
"trial_name": null,
"trial_params": null
}