m2m100_1.2B_ft_kbd-ru_63K / train_results.json
anzorq's picture
End of training
824dc30
raw
history blame
203 Bytes
{
"epoch": 2.27,
"train_loss": 5.4469778463318936e-05,
"train_runtime": 1.6018,
"train_samples": 63407,
"train_samples_per_second": 89898.478,
"train_steps_per_second": 5618.655
}