andrecastro's picture
End of training
b51ea79
raw
history blame
209 Bytes
{
"epoch": 3.0,
"total_flos": 1.8113127425839596e+18,
"train_loss": 0.2791935749937685,
"train_runtime": 1159.1194,
"train_samples_per_second": 62.893,
"train_steps_per_second": 1.964
}