UserGroup3-llava / trainer_state.json
oaishi's picture
Add SFT checkpoint (LLaVA-Next 8B on CowCorpus)
bff9fb4 verified
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 34,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.058823529411764705,
"grad_norm": 19.060205459594727,
"learning_rate": 2.5e-08,
"loss": 0.1856,
"num_input_tokens_seen": 30264,
"step": 1
},
{
"epoch": 0.5882352941176471,
"grad_norm": 15.253729820251465,
"learning_rate": 9.045084971874737e-08,
"loss": 0.1248,
"num_input_tokens_seen": 288472,
"step": 10
},
{
"epoch": 1.1764705882352942,
"grad_norm": 12.735420227050781,
"learning_rate": 4.477357683661733e-08,
"loss": 0.1341,
"num_input_tokens_seen": 575560,
"step": 20
},
{
"epoch": 1.7647058823529411,
"grad_norm": 7.144538402557373,
"learning_rate": 4.322727117869951e-09,
"loss": 0.1508,
"num_input_tokens_seen": 869208,
"step": 30
},
{
"epoch": 2.0,
"num_input_tokens_seen": 977792,
"step": 34,
"total_flos": 5267146031104.0,
"train_loss": 0.12937388700597427,
"train_runtime": 896.4768,
"train_samples_per_second": 0.303,
"train_steps_per_second": 0.038
}
],
"logging_steps": 10,
"max_steps": 34,
"num_input_tokens_seen": 977792,
"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": 5267146031104.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}