hp_ablations_mistral_bsz256 / train_results.json
sedrickkeh's picture
End of training
1874780 verified
raw
history blame contribute delete
221 Bytes
{
"epoch": 2.9984810126582278,
"total_flos": 2479683262218240.0,
"train_loss": 0.46617310421968466,
"train_runtime": 84499.4991,
"train_samples_per_second": 8.974,
"train_steps_per_second": 0.035
}