Summary of Augmenting Math Word Problems Via Iterative Question Composing, by Haoxiong Liu et al.
Augmenting Math Word Problems via Iterative Question Composing
by Haoxiong Liu, Yifan Zhang, Yifan Luo, Andrew Chi-Chih Yao
First submitted to arxiv on: 17 Jan 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces the MMIQC dataset, designed to enhance the mathematical reasoning capabilities of base language models. The dataset combines processed web data and synthetic question-response pairs. Models fine-tuned on MMIQC outperform their counterparts on the MATH benchmark across various model sizes, with Qwen-72B-MMIQC achieving a 45.0% accuracy, surpassing the open-source state-of-the-art by 8.2%. The improvement generalizes to unseen data, as demonstrated through evaluation results on Hungarian high school finals. The authors also conduct an ablation study, attributing a large part of the improvement to their novel augmentation method, Iterative Question Composing (IQC). |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper helps language models get better at math problems. It makes a special dataset called MMIQC that combines internet data and questions that were made artificially. When language models are trained on this new dataset, they can solve math problems much better than before! The best model got 45% of the answers right, which is 8% better than what other models could do. This means it’s not just good at solving one type of problem, but also works well with new math questions. |
