andrecastro's picture
End of training
21de218
raw
history blame
209 Bytes
{
"epoch": 2.99,
"total_flos": 3.815394309328896e+17,
"train_loss": 0.09319638077480098,
"train_runtime": 491.2433,
"train_samples_per_second": 31.384,
"train_steps_per_second": 0.977
}