Model-J__MAE__model_idx_0202 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
fb1de49 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.8810666666666667,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0202/checkpoints/checkpoint-1332",
"epoch": 4.0,
"eval_steps": 500,
"global_step": 1332,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"grad_norm": 4.903066158294678,
"learning_rate": 5e-05,
"loss": 1.9077,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.8026666666666666,
"eval_loss": 0.7685175538063049,
"eval_runtime": 5.9617,
"eval_samples_per_second": 629.02,
"eval_steps_per_second": 9.897,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 1.0240625143051147,
"learning_rate": 5e-05,
"loss": 0.5223,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8541333333333333,
"eval_loss": 0.5066100358963013,
"eval_runtime": 6.3439,
"eval_samples_per_second": 591.122,
"eval_steps_per_second": 9.3,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": Infinity,
"learning_rate": 5e-05,
"loss": 0.2896,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.8808,
"eval_loss": 0.38548117876052856,
"eval_runtime": 6.3433,
"eval_samples_per_second": 591.176,
"eval_steps_per_second": 9.301,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 0.403563529253006,
"learning_rate": 5e-05,
"loss": 0.1757,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.8810666666666667,
"eval_loss": 0.3954995572566986,
"eval_runtime": 5.9472,
"eval_samples_per_second": 630.554,
"eval_steps_per_second": 9.921,
"step": 1332
}
],
"logging_steps": 500,
"max_steps": 1665,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 6.58965282490368e+18,
"train_batch_size": 64,
"trial_name": null,
"trial_params": null
}