Qwen2-72B-SFT-Step-DPO / train_results.json
xinlai's picture
upload model
e350c62
{
"epoch": 3.982222222222222,
"total_flos": 0.0,
"train_loss": 0.19470643034825721,
"train_runtime": 59934.0013,
"train_samples": 10795,
"train_samples_per_second": 0.72,
"train_steps_per_second": 0.006
}