Qwen2-72B-Instruct-Step-DPO / train_results.json
xinlai's picture
upload model
ee76254
{
"epoch": 3.982222222222222,
"total_flos": 0.0,
"train_loss": 0.14947671072912358,
"train_runtime": 64472.8667,
"train_samples": 10795,
"train_samples_per_second": 0.67,
"train_steps_per_second": 0.005
}