100 lines
2.4 KiB
JSON
100 lines
2.4 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.201765447667087,
|
|
"eval_steps": 500,
|
|
"global_step": 40,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.025220680958385876,
|
|
"grad_norm": 13.8125,
|
|
"learning_rate": 7.883767269704208e-07,
|
|
"loss": 1.7700687408447267,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.05044136191677175,
|
|
"grad_norm": 14.0625,
|
|
"learning_rate": 7.197771053614005e-07,
|
|
"loss": 1.7459712982177735,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.07566204287515763,
|
|
"grad_norm": 13.875,
|
|
"learning_rate": 6e-07,
|
|
"loss": 1.5911680221557618,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.1008827238335435,
|
|
"grad_norm": 15.5625,
|
|
"learning_rate": 4.4821467210212917e-07,
|
|
"loss": 1.718311882019043,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.12610340479192939,
|
|
"grad_norm": 14.5625,
|
|
"learning_rate": 2.8871301443341895e-07,
|
|
"loss": 1.7665422439575196,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.15132408575031525,
|
|
"grad_norm": 12.625,
|
|
"learning_rate": 1.4702184976184913e-07,
|
|
"loss": 1.7348012924194336,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.17654476670870115,
|
|
"grad_norm": 14.6875,
|
|
"learning_rate": 4.581758973871608e-08,
|
|
"loss": 1.7532463073730469,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.201765447667087,
|
|
"grad_norm": 14.0625,
|
|
"learning_rate": 1.2970767463160281e-09,
|
|
"loss": 1.7677625656127929,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.201765447667087,
|
|
"step": 40,
|
|
"total_flos": 1.92457717123584e+16,
|
|
"train_loss": 1.7309840440750122,
|
|
"train_runtime": 323.247,
|
|
"train_samples_per_second": 0.491,
|
|
"train_steps_per_second": 0.124
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 40,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": false,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1.92457717123584e+16,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|