Summary of Can Small Language Models Help Large Language Models Reason Better?: Lm-guided Chain-of-thought, by Jooyoung Lee et al.
Can Small Language Models Help Large Language Models Reason Better?: LM-Guided Chain-of-Thought
by Jooyoung Lee, Fan Yang, Thanh Tran, Qian Hu, Emre Barut, Kai-Wei Chang, Chengwei Su
First submitted to arxiv on: 4 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
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 We propose a novel framework, LM-Guided CoT, which leverages a lightweight language model (LM) to guide a large LM in reasoning tasks. The lightweight LM generates a rationale for each input instance, and the large LM predicts a task output based on this rationale. Our approach is resource-efficient, requiring only the training of the lightweight LM. We optimize the model through knowledge distillation and reinforcement learning from rationale-oriented and task-oriented reward signals. Our method outperforms baselines in multi-hop extractive question answering (QA) benchmarks like HotpotQA and 2WikiMultiHopQA, with improved answer prediction accuracy. Additionally, we find that reinforcement learning improves the quality of rationales and QA performance. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary We’ve developed a new way to help computers understand language better. Our method uses a small computer program (called a lightweight model) to help a bigger computer program make decisions about what answer is correct. The small model explains why it thinks one answer is right, and the big model uses this explanation to choose an answer. We tested our approach on hard questions that require understanding lots of information, and it worked better than other methods. Our method also helps computers come up with better explanations for their answers. |
Keywords
» Artificial intelligence » Knowledge distillation » Language model » Question answering » Reinforcement learning
