Qwen2.5-3B-Instruct-RG-Math / all_results.json
Zafir Stojanovski
Add checkpoint 600 post-trained on curated_rg_math
1384542 verified
{
"total_flos": 0.0,
"train_loss": 0.15968478212249465,
"train_runtime": 15013.7279,
"train_samples_per_second": 0.639,
"train_steps_per_second": 0.04
}