Qwen2.5-Math-7B-Instruct-DAPO-G8 / train_results.json
JH Na
Model save
03f3366 verified
{
"total_flos": 0.0,
"train_loss": 0.002357518015606342,
"train_runtime": 35919.5545,
"train_samples": 5580,
"train_samples_per_second": 0.777,
"train_steps_per_second": 0.024
}