Summary of From Explicit Cot to Implicit Cot: Learning to Internalize Cot Step by Step, By Yuntian Deng et al.
From Explicit CoT to Implicit CoT: Learning to Internalize CoT Step by Step
by Yuntian Deng, Yejin Choi, Stuart Shieber
First submitted to arxiv on: 23 May 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 A novel approach for internalizing chain-of-thought (CoT) steps in language models is proposed to achieve high accuracy in reasoning tasks. By starting with a model trained for explicit CoT reasoning and gradually removing intermediate steps, the model learns to simplify the reasoning process while maintaining performance. This method enables a GPT-2 Small model to solve 9-by-9 multiplication with up to 99% accuracy, surpassing standard training’s limitations of 4-by-4 multiplication. The approach also proves effective on larger language models like Mistral 7B, achieving over 50% accuracy on GSM8K without producing intermediate steps. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to help computers understand how to solve problems is discovered. It starts with a computer program that explains its thinking step-by-step and then removes those extra steps. This helps the program simplify its thinking while still getting answers right. With this method, a small computer program can now solve multiplication problems as big as 9 x 9 with almost perfect accuracy! The same approach works on bigger programs too. |
Keywords
» Artificial intelligence » Gpt
