Llama-3.1-8B-Instruct-SAA-Half / train_results.json
chchen's picture
End of training
b3ab670 verified
raw
history blame contribute delete
218 Bytes
{
"epoch": 2.986666666666667,
"total_flos": 1.516498186272768e+16,
"train_loss": 1.23334006468455,
"train_runtime": 194.0073,
"train_samples_per_second": 6.959,
"train_steps_per_second": 0.433
}