---
datasets:
- openai/gsm8k
base_model:
- Qwen/Qwen2.5-3B-Instruct
---

System prompt:
```
Respond in the following format:
<reasoning>
...
</reasoning>
<answer>
...
</answer>
```