m2m100_418M_fr_wol_rel / train_results.json
Davlan's picture
add MT model
b1767f2
{
"epoch": 3.0,
"train_loss": 1.9876760678924648,
"train_runtime": 2038.2062,
"train_samples": 22002,
"train_samples_per_second": 32.384,
"train_steps_per_second": 3.24
}