bert_12_layer_model_v2 / train_results.json
gokuls's picture
End of training
810ced4
{
"epoch": 5.0,
"train_loss": 3.027850917318448,
"train_runtime": 97726.2074,
"train_samples": 2929379,
"train_samples_per_second": 149.877,
"train_steps_per_second": 2.342
}