Summary of Dotamath: Decomposition Of Thought with Code Assistance and Self-correction For Mathematical Reasoning, by Chengpeng Li et al.
DotaMath: Decomposition of Thought with Code Assistance and Self-correction for Mathematical Reasoning
by Chengpeng Li, Guanting Dong, Mingfeng Xue, Ru Peng, Xiang Wang, Dayiheng Liu
First submitted to arxiv on: 4 Jul 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 a series of Large Language Models (LLMs) that employ the Decomposition of thought with code assistance and self-correction for mathematical reasoning, dubbed as DotaMath. The models tackle complex mathematical tasks by decomposing them into simpler logical subtasks, leveraging code to solve these subtasks, obtaining fine-grained feedback from the code interpreter, and engaging in self-reflection and correction. The paper generates an instruction fine-tuning dataset called DotaMathQA with 574K query-response pairs using diverse interactive tool-use trajectories and employing query evolution on GSM8K and MATH datasets. The trained LLMs achieve remarkable performance compared to open-source LLMs across various in-domain and out-of-domain benchmarks, showcasing exceptional results of 64.8% on the competitive MATH dataset and 86.7% on GSM8K with DotaMath-deepseek-7B. Additionally, the paper presents strong competitiveness on a series of in-domain and out-of-domain benchmarks (Avg. 80.1%). The authors anticipate that the DotaMath paradigm will open new pathways for addressing intricate mathematical problems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper introduces a way to help computers understand complex math problems better. They created a system called DotaMath, which breaks down difficult math tasks into smaller steps and uses code to solve those steps. This process helps the computer learn from its mistakes and improve over time. The researchers also developed a large dataset of math questions and answers that they used to train their models. The results show that DotaMath can understand math problems better than other systems, with some models achieving scores of 64.8% on one test and 86.7% on another. This could lead to new ways for computers to solve complex math problems in the future. |
Keywords
* Artificial intelligence * Fine tuning
