jkson's picture
End of training
13ea625
raw
history blame contribute delete
208 Bytes
{
"epoch": 3.0,
"total_flos": 1.8124066505760768e+18,
"train_loss": 0.3384138770270766,
"train_runtime": 1144.997,
"train_samples_per_second": 63.668,
"train_steps_per_second": 0.498
}