RobertoSonic's picture
End of training
b82012d verified
{
"epoch": 32.0,
"total_flos": 5.861874920136376e+17,
"train_loss": 0.7368841990828514,
"train_runtime": 406.4801,
"train_samples_per_second": 55.402,
"train_steps_per_second": 0.394
}