Summary of Enhancing Llm Reasoning Via Critique Models with Test-time and Training-time Supervision, by Zhiheng Xi et al.
Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision
by Zhiheng Xi, Dingwen Yang, Jixuan Huang, Jiafu Tang, Guanyu Li, Yiwen Ding, Wei He, Boyang Hong, Shihan Do, Wenyu Zhan, Xiao Wang, Rui Zheng, Tao Ji, Xiaowei Shi, Yitao Zhai, Rongxiang Weng, Jingang Wang, Xunliang Cai, Tao Gui, Zuxuan Wu, Qi Zhang, Xipeng Qiu, Xuanjing Huang, Yu-Gang Jiang
First submitted to arxiv on: 25 Nov 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 proposes a two-player paradigm for training large language models (LLMs) to reason mathematically. The critique model provides step-level feedback to supervise the reasoning model during both test-time and train-time. The authors introduce AutoMathCritique, a framework for collecting critique data, resulting in a dataset of 76,321 responses paired with step-level feedback. Fine-tuning LLMs with this dataset enables them to generate natural language feedback for mathematical reasoning. The paper shows that the critique models consistently improve the actor’s performance on difficult queries at test-time, especially when scaling up inference-time computation. The authors also propose a critique-in-the-loop self-improvement method, which improves the actor’s exploration efficiency and solution diversity, especially on challenging queries. The preliminary step to explore training self-talk reasoning models via critique supervision is showcased. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps machines learn to solve math problems better by giving them feedback while they think. Usually, these machines just spit out an answer without thinking about it first. But this paper shows that if we give them feedback at each step, they’ll become much better problem-solvers. The researchers created a special dataset with lots of math problems and the correct answers to help train these machines. They also developed a new way for these machines to learn from their mistakes and improve over time. This could be useful in fields like science, coding, and mathematics. |
Keywords
» Artificial intelligence » Fine tuning » Inference
