### GSM8K-Zero
This dataset is constructed from GSM8K to test the over-reasoning and redundant calculation in the responses of LLMs.
For more details, please refer to [our github repo](https://github.com/d223302/Over-Reasoning-of-LLMs) or our [EACL'24 paper](https://openreview.net/forum?id=Glx1e3bYU5).