Summary of Llm2: Let Large Language Models Harness System 2 Reasoning, by Cheng Yang et al.
LLM2: Let Large Language Models Harness System 2 Reasoning
by Cheng Yang, Chufan Shi, Siheng Li, Bo Shui, Yujiu Yang, Wai Lam
First submitted to arxiv on: 29 Dec 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 The proposed framework, LLM2, combines a large language model (LLM) with a process-based verifier to mitigate undesirable outputs. By introducing dual-process theory-inspired mechanisms, LLM2 generates plausible candidates while the verifier provides feedback to distinguish desirable and undesirable results. The verifier is trained on synthetic data using a pairwise comparison loss, achieving accuracy enhancements of up to +14.0 on mathematical reasoning benchmarks like GSM8K. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large language models are getting better at many tasks, but sometimes they make mistakes. To fix this, researchers created a new way to combine two types of thinking: the first is creative and generates ideas, while the second checks if those ideas are good or bad. This system is called LLM2. It’s like having a friend who helps you decide if your ideas are worth sharing. The new system did better on math problems than previous ones, which is very important for making sure we can rely on these language models. |
Keywords
» Artificial intelligence » Large language model » Synthetic data
