Mistral-7B-Instruct-v0.2-ORPO-SFT / train_results.json
chchen's picture
End of training
f09da46 verified
raw
history blame contribute delete
222 Bytes
{
"epoch": 2.986666666666667,
"total_flos": 1.6019812625547264e+16,
"train_loss": 0.12199947628236953,
"train_runtime": 363.0569,
"train_samples_per_second": 7.437,
"train_steps_per_second": 0.463
}