llama-7b-dpo-qlora-relu / train_results.json
chanchan7's picture
Model save
7221958 verified
{
"epoch": 1.0,
"train_loss": 0.6582325137556925,
"train_runtime": 111789.0929,
"train_samples": 61135,
"train_samples_per_second": 0.547,
"train_steps_per_second": 0.034
}