byt5_fr_ewe_news / trainer_state.json
Davlan's picture
add MT model
0fa6efa
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 10.0,
"global_step": 2030,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 2.46,
"learning_rate": 3.768472906403941e-05,
"loss": 1.9158,
"step": 500
},
{
"epoch": 4.93,
"learning_rate": 2.5369458128078822e-05,
"loss": 1.5117,
"step": 1000
},
{
"epoch": 7.39,
"learning_rate": 1.3054187192118228e-05,
"loss": 1.408,
"step": 1500
},
{
"epoch": 9.85,
"learning_rate": 7.389162561576355e-07,
"loss": 1.3623,
"step": 2000
},
{
"epoch": 10.0,
"step": 2030,
"total_flos": 2.417011001553715e+16,
"train_loss": 1.5465415597549212,
"train_runtime": 3081.6278,
"train_samples_per_second": 6.574,
"train_steps_per_second": 0.659
}
],
"max_steps": 2030,
"num_train_epochs": 10,
"total_flos": 2.417011001553715e+16,
"trial_name": null,
"trial_params": null
}