The Orca-Math Elementary Math Problem Dataset is a dataset containing approximately 200,000 elementary school math word problems, released by Microsoft in 2024. The related paper is "Orca-Math: Unlocking the potential of SLMs in Grade School Math", aiming to enhance the mathematical capabilities of language models and provide a solid foundation for mathematical problem solving.

The dataset uses Azure GPT-4-Turbo to generate solutions to the math problems, covering various types of elementary school math word problems. It can be used for mathematical instruction fine-tuning, synthetic data augmentation, and research on model mathematical reasoning capabilities.

Dataset Composition

The dataset contains a training set (train) with a total of 200,035 samples. Each sample includes the following two main fields:

• question: English math word problem.
• answer: Solution to the corresponding question.

Citation

@misc{mitra2024orcamath,
      title={Orca-Math: Unlocking the potential of SLMs in Grade School Math}, 
      author={Arindam Mitra and Hamed Khanpour and Corby Rosset and Ahmed Awadallah},
      year={2024},
      eprint={2402.14830},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}