karoladelk's picture
End of training
dc461d0 verified
raw
history blame
210 Bytes
{
"epoch": 9.98,
"total_flos": 6.364199987970048e+18,
"train_loss": 0.11027517792582511,
"train_runtime": 4749.2843,
"train_samples_per_second": 54.038,
"train_steps_per_second": 0.211
}