phi-2-gpo-test-longest-iter-2 / train_results.json
BraylonDash's picture
Model save
266f7bc verified
{
"epoch": 1.98,
"train_loss": 0.0102488252845022,
"train_runtime": 3093.4739,
"train_samples": 61135,
"train_samples_per_second": 0.647,
"train_steps_per_second": 0.04
}