---
datasets:
- meta-math/MetaMathQA
---

## Introduction
The model is trained with Masked Thought Fine-Tuning (MFT), a simple variant of standard Supervised Fine-Tuning (SFT). You can refer to our code and paper below.

## Links
- **Code**: [https://github.com/ChangyuChen347/MaskedThought](https://github.com/ChangyuChen347/MaskedThought)
- **Paper**: [https://arxiv.org/abs/2403.02178](https://arxiv.org/abs/2403.02178)

## Results
We test it with the scripts provided in [MetaMath](https://github.com/meta-math/MetaMath).
| Model                                                                                                                                 | GSM8K | MATH  |
|--------------------------------------------------------------------------------------------------------------------------------------------------|-------|-------|
| [adalaw/MetaMath-Mistral-7B-MFT](https://huggingface.co/adalaw/MetaMath-Mistral-7B-MFT)                                                          | 79.90 | 29.0  |
| [meta-math/MetaMath-Mistral-7B-SFT](https://huggingface.co/meta-math/MetaMath-Mistral-7B)                                                        | 77.70 | 28.2  |