UserGroup1-llava / trainer_state.json
oaishi's picture
Add SFT checkpoint (LLaVA-Next 8B on CowCorpus)
8b29390 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9454545454545453,
"eval_steps": 500,
"global_step": 54,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03636363636363636,
"grad_norm": 40.4747428894043,
"learning_rate": 1.6666666666666664e-08,
"loss": 1.5645,
"num_input_tokens_seen": 31688,
"step": 1
},
{
"epoch": 0.36363636363636365,
"grad_norm": 39.040138244628906,
"learning_rate": 9.82962913144534e-08,
"loss": 1.5554,
"num_input_tokens_seen": 310296,
"step": 10
},
{
"epoch": 0.7272727272727273,
"grad_norm": 44.81167984008789,
"learning_rate": 8.043807145043603e-08,
"loss": 1.5891,
"num_input_tokens_seen": 623472,
"step": 20
},
{
"epoch": 1.0727272727272728,
"grad_norm": 34.17668151855469,
"learning_rate": 5e-08,
"loss": 1.4807,
"num_input_tokens_seen": 914512,
"step": 30
},
{
"epoch": 1.4363636363636363,
"grad_norm": 35.74896240234375,
"learning_rate": 1.9561928549563966e-08,
"loss": 1.5438,
"num_input_tokens_seen": 1223320,
"step": 40
},
{
"epoch": 1.8,
"grad_norm": 39.04248046875,
"learning_rate": 1.70370868554659e-09,
"loss": 1.5575,
"num_input_tokens_seen": 1535600,
"step": 50
},
{
"epoch": 1.9454545454545453,
"num_input_tokens_seen": 1659024,
"step": 54,
"total_flos": 8930251325440.0,
"train_loss": 1.5560528790509258,
"train_runtime": 1936.0529,
"train_samples_per_second": 0.226,
"train_steps_per_second": 0.028
}
],
"logging_steps": 10,
"max_steps": 54,
"num_input_tokens_seen": 1659024,
"num_train_epochs": 2,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 8930251325440.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}