Skip to content

Latest commit

 

History

History
46 lines (25 loc) · 1.98 KB

README.md

File metadata and controls

46 lines (25 loc) · 1.98 KB
Raw

Math Problem Solver

This repository contains a deep learning-based model for solving various types of math problems.

Overview

The purpose of this project is to develop a robust model capable of solving a wide range of mathematical problems using deep learning techniques. We initially experimented with several open-source models to establish a baseline for performance. Subsequently, we fine-tuned the DeepSeek-Math model with Qlora to enhance its capabilities and accuracy.

Features

  • Solves diverse types of math problems.
  • Utilizes state-of-the-art deep learning techniques.
  • Easy integration into existing systems or applications.

Usage

To use the fine-tuned DeepSeek-Math model with Qlora, download the Lora adapters from Kaggle and follow the instructions provided in the model card for integration.

Models Used

  • Open-Source Models: We experimented with various open-source models available in the literature.
  • Fine-Tuned Model: The DeepSeek-Math model was fine-tuned with Qlora to improve its accuracy and performance. For a detailed fine-tuning report, refer to W&B report.

Fine-Tuning Report

Fine-Tuning Report 1

Fine-Tuning Report 2

Data

Data used for training and fine-tuning the models are not included in this repository due to licensing restrictions. Instructions for obtaining the data can be found here.

Acknowledgements

  • We thank the developers of the open-source models used in this project for their contributions.
  • Special thanks to the team behind Qlora for providing a powerful platform for fine-tuning deep learning models.

License

This project is licensed under the MIT License.

Contributing

Contributions are welcome! Please feel free to open issues or pull requests.