Summary of Probing Language Models For Pre-training Data Detection, by Zhenhua Liu et al.

Main: Computation and Language (cs.CL)
Secondary: Artificial Intelligence (cs.AI)

Probing Language Models for Pre-training Data Detection

by Zhenhua Liu, Tong Zhu, Chuanyuan Tan, Haonan Lu, Bing Liu, Wenliang Chen

First submitted to arxiv on: 3 Jun 2024

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 method uses probing technique to detect pre-training data contamination in Large Language Models (LLMs) by examining the model’s internal activations. This approach outperforms baselines and achieves state-of-the-art performance on WikiMIA and ArxivMIA benchmarks, demonstrating its effectiveness. The study also introduces ArxivMIA, a new challenging benchmark comprising arxiv abstracts from Computer Science and Mathematics categories.
Low	GrooveSquid.com (original content)	Low Difficulty Summary The researchers created a way to check if Large Language Models (LLMs) have been trained using certain data by looking at the model’s internal workings. This method is simple and good at finding contaminated data. It did better than other methods and even works well with new datasets.