byt5_mos_fr_news / trainer_state.json
Davlan's picture
add MT model
02892b5
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 10.0,
"global_step": 2290,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 2.18,
"learning_rate": 3.9082969432314415e-05,
"loss": 1.309,
"step": 500
},
{
"epoch": 4.37,
"learning_rate": 2.816593886462882e-05,
"loss": 1.053,
"step": 1000
},
{
"epoch": 6.55,
"learning_rate": 1.7248908296943234e-05,
"loss": 0.9836,
"step": 1500
},
{
"epoch": 8.73,
"learning_rate": 6.3318777292576415e-06,
"loss": 0.9428,
"step": 2000
},
{
"epoch": 10.0,
"step": 2290,
"total_flos": 3.1800277338660864e+16,
"train_loss": 1.0541186869925288,
"train_runtime": 1883.4725,
"train_samples_per_second": 12.142,
"train_steps_per_second": 1.216
}
],
"max_steps": 2290,
"num_train_epochs": 10,
"total_flos": 3.1800277338660864e+16,
"trial_name": null,
"trial_params": null
}