m2m100_418M_en_swa_rel / train_results.json
Davlan's picture
add MT model
fe30341
{
"epoch": 3.0,
"train_loss": 0.8388812589635725,
"train_runtime": 66410.1599,
"train_samples": 872008,
"train_samples_per_second": 39.392,
"train_steps_per_second": 3.939
}