djbp's picture
End of training
bccc1b3 verified
raw
history blame
202 Bytes
{
"epoch": 3.0,
"total_flos": 3525591661885440.0,
"train_loss": 0.2555307348569234,
"train_runtime": 72.9305,
"train_samples_per_second": 0.617,
"train_steps_per_second": 0.041
}