Llama-3.1-8B-Instruct-SAA-800 / train_results.json
chchen's picture
End of training
d7d433d verified
raw
history blame contribute delete
207 Bytes
{
"epoch": 10.0,
"total_flos": 8.074648418018918e+16,
"train_loss": 0.3428536836306254,
"train_runtime": 994.0751,
"train_samples_per_second": 7.243,
"train_steps_per_second": 0.453
}