hp_ablations_gemma_bsz512 / train_results.json
sedrickkeh's picture
End of training
482ac3e verified
raw
history blame contribute delete
220 Bytes
{
"epoch": 2.9991537376586743,
"total_flos": 5064195066298368.0,
"train_loss": 0.5477321235823218,
"train_runtime": 121628.759,
"train_samples_per_second": 5.596,
"train_steps_per_second": 0.011
}