Summary of Pico: Peer Review in Llms Based on the Consistency Optimization, by Kun-peng Ning et al.
PiCO: Peer Review in LLMs based on the Consistency Optimization
by Kun-Peng Ning, Shuo Yang, Yu-Yang Liu, Jia-Yu Yao, Zhen-Hui Liu, Yong-Hong Tian, Yibing Song, Li Yuan
First submitted to arxiv on: 2 Feb 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 This paper proposes a novel unsupervised evaluation method for large language models (LLMs) using peer-review mechanisms. The approach allows LLMs to evaluate each other’s responses without human annotations, and assigns a learnable capability parameter to determine the ability hierarchy among models. The goal is to maximize consistency between model capabilities and scores. Three metrics – PEN, CIN, and LIS – are introduced to measure the gap in aligning human rankings. Experimental results on multiple datasets validate the effectiveness of this approach. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper finds a new way to test big language models without human help. It lets these models evaluate each other’s answers and ranks them based on how well they do. The idea is that high-level models are better at judging lower-level models’ responses, and can also do well themselves. The authors come up with three ways to measure how good this system is. They test it on different datasets and show that it works. |
Keywords
* Artificial intelligence * Unsupervised
