Summary of Self-play with Execution Feedback: Improving Instruction-following Capabilities Of Large Language Models, by Guanting Dong et al.
Self-play with Execution Feedback: Improving Instruction-following Capabilities of Large Language Models
by Guanting Dong, Keming Lu, Chengpeng Li, Tingyu Xia, Bowen Yu, Chang Zhou, Jingren Zhou
First submitted to arxiv on: 19 Jun 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 AutoIF, a scalable and reliable method for automatically generating instruction-following training data to enhance the complex instruction-following abilities of large language models (LLMs). The approach transforms validation into code verification, requiring LLMs to generate instructions, corresponding code, and unit test samples. Execution feedback-based rejection sampling generates data for Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) training. AutoIF achieves significant improvements across three training algorithms when applied to top open-source LLMs Qwen2 and LLaMA3 in self-alignment and strong-to-weak distillation settings. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper creates a way to teach large language models how to follow instructions. Right now, it’s hard to make good data for this task without putting in a lot of human effort. The new method, called AutoIF, makes it easier by turning the process into a code-checking problem. This helps create high-quality training data that can be used with different teaching methods like supervised fine-tuning and reinforcement learning from human feedback. When tested on two popular language models, AutoIF showed big improvements. |
Keywords
» Artificial intelligence » Alignment » Distillation » Fine tuning » Reinforcement learning from human feedback » Rlhf » Supervised
