Llama-3.1-8B-Instruct-SFT / train_results.json
dslighfdsl's picture
Model save
322ab5a verified
{
"total_flos": 1.1600758474539008e+16,
"train_loss": 0.5818708042303721,
"train_runtime": 635.0171,
"train_samples": 1483,
"train_samples_per_second": 0.454,
"train_steps_per_second": 0.028
}