alex2awesome's picture
model trained with deepspeed 2
bf57257
raw
history blame contribute delete
193 Bytes
{
"epoch": 3.0,
"train_loss": 1.3908403631165627,
"train_runtime": 5877.8179,
"train_samples": 8277,
"train_samples_per_second": 4.225,
"train_steps_per_second": 2.113
}