Measuring Fingerprints of Web-filtered Text Datasets and Fingerprint Propagation Through Training

Youssef Mansour, Reinhard Heckel

Keywords: fingerprints, text datasets, propagation, generalisation, mixture proportions, finetuning sources

TL;DR: detectibility of fingerprints in text datasets and their propagation and implications

Abstract: We investigate fingerprints in pretraining datasets for large language models (LLMs) through dataset classification experiments. Building on prior work demonstrating the existence of fingerprints or biases in popular computer vision datasets, we analyze popular open-source pretraining datasets for LLMs derived from CommonCrawl including C4, RefinedWeb, DolmaCC, RedPajama-V2, FineWeb, and DCLM-Baseline. Despite those datasets being obtained with similar curation steps, neural networks can classify surprisingly well which dataset a single text sequence belongs to, significantly better than a human can. This indicates that small differences in filtering and processing pipelines induce fingerprints, that we find are evident in formatting, vocabulary, and content distributions. Such fingerprints can negatively impact cross-dataset generalization. Additionally, we show that these fingerprints propagate through training: sequences generated by models trained on those datasets can be accurately classified by a classifier trained on the original datasets. This can offer insights into data characteristics that are typically undisclosed by LLM developers, including pretraining mixture proportions and finetuning data sources.

Code URL: https://github.com/MLI-lab/LLM_data_bias

Primary Area: Datasets & Benchmarks illustrating Different Deep learning Scenarios (e.g., architectures, generative models, optimization for deep networks, foundation models, LLMs)

Submission Number: 2124