76 lines
1.9 KiB
JSON
76 lines
1.9 KiB
JSON
|
{
|
||
|
|
"best_global_step": 1268,
|
||
|
|
"best_metric": 69.08031412631969,
|
||
|
|
"best_model_checkpoint": "mbert-loraxs-qa-vanilla/run-5/checkpoint-1268",
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 1268,
|
||
|
|
"is_hyper_param_search": true,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"grad_norm": 54.67623519897461,
|
||
|
|
"learning_rate": 0.00019176112053781796,
|
||
|
|
"loss": 2.763942008514886,
|
||
|
|
"step": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_exact_match": 51.68,
|
||
|
|
"eval_f1": 65.72042360682833,
|
||
|
|
"eval_loss": 1.602353572845459,
|
||
|
|
"eval_runtime": 7.6892,
|
||
|
|
"eval_samples_per_second": 164.517,
|
||
|
|
"eval_steps_per_second": 10.404,
|
||
|
|
"step": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"grad_norm": 43.33531188964844,
|
||
|
|
"learning_rate": 0.0001438397368480361,
|
||
|
|
"loss": 1.6214439801237184,
|
||
|
|
"step": 1268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_exact_match": 55.12,
|
||
|
|
"eval_f1": 69.08031412631969,
|
||
|
|
"eval_loss": 1.4728764295578003,
|
||
|
|
"eval_runtime": 7.7275,
|
||
|
|
"eval_samples_per_second": 163.7,
|
||
|
|
"eval_steps_per_second": 10.353,
|
||
|
|
"step": 1268
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 500,
|
||
|
|
"max_steps": 3170,
|
||
|
|
"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": 5023145924278272.0,
|
||
|
|
"train_batch_size": 16,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": {
|
||
|
|
"learning_rate": 0.0002156462266040184,
|
||
|
|
"lora_alpha": 512,
|
||
|
|
"lora_dropout": 0.085847996108665,
|
||
|
|
"lora_r": 128,
|
||
|
|
"num_train_epochs": 5,
|
||
|
|
"weight_decay": 0.06498728377281919
|
||
|
|
}
|
||
|
|
}
|