Isaacks's picture
End of training
acd0154
{
"epoch": 3.0,
"total_flos": 1.2401345913255936e+17,
"train_loss": 0.33031844481443745,
"train_runtime": 195.8421,
"train_samples_per_second": 25.475,
"train_steps_per_second": 0.199
}