Model-J__DINO__model_idx_0178 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
9d9f3d6 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.2693333333333333,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0178/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": Infinity,
"learning_rate": 0.0005,
"loss": 3.443,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.16106666666666666,
"eval_loss": 3.1117987632751465,
"eval_runtime": 12.4012,
"eval_samples_per_second": 302.39,
"eval_steps_per_second": 4.758,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 19.803543090820312,
"learning_rate": 0.0005,
"loss": 3.0021,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.204,
"eval_loss": 2.989069700241089,
"eval_runtime": 12.4355,
"eval_samples_per_second": 301.556,
"eval_steps_per_second": 4.744,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 12.614653587341309,
"learning_rate": 0.0005,
"loss": 2.8497,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.2504,
"eval_loss": 2.7954177856445312,
"eval_runtime": 12.4301,
"eval_samples_per_second": 301.687,
"eval_steps_per_second": 4.747,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 17.939966201782227,
"learning_rate": 0.0005,
"loss": 2.7543,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.2693333333333333,
"eval_loss": 2.719097375869751,
"eval_runtime": 12.3781,
"eval_samples_per_second": 302.955,
"eval_steps_per_second": 4.766,
"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
}