jemole's picture
End of training
1748811
raw
history blame contribute delete
209 Bytes
{
"epoch": 3.0,
"total_flos": 1.8124066505760768e+18,
"train_loss": 0.3311593678959629,
"train_runtime": 2581.5594,
"train_samples_per_second": 28.239,
"train_steps_per_second": 0.221
}