Data Portraits: Recording Foundation Model Training Data

By Marc Marone, Benjamin Van Durme. 2023

43 citations recorded by the source corpus.

Foundation models are trained on increasingly immense and opaque datasets. Even while these models are now key in AI system building, it can be difficult to answer the straightforward question: has the model already encountered a given example during training? We therefore propose a widespread adoption of Data Portraits: artifacts that record training data and allow for downstream inspection. First we outline the properties of such an artifact and discuss how existing solutions can be used to increase transparency. We then propose and implement a solution based on data sketching, stressing fast and space efficient querying. Using our tools, we document a popular language modeling corpus (The Pile) and a recently released code modeling dataset (The Stack). We show that our solution enables answering questions about test set leakage and model plagiarism. Our tool is lightweight and fast, costing only 3% of the dataset size in overhead. We release a live interface of our tools at https://dataportraits.org/ and call on dataset and model creators to release Data Portraits as a complement to current documentation practices.

Subfields: Large Language Models, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Large Language Models

In Evaluation & Benchmarks

Citations in the corpus

StarCoder: may the source be with you! (2023, background): “This pressing need for understanding data (Mitchell et al., 2022) is being increasingly recognized and operationalized in the form of dataset inspection tools and toolkits (Akiki et al., 2023; Marone & Van Durme, 2023; Piktus et al., 2023).”
NLP Evaluation in trouble: On the Need to Measure LLM Data Contamination for each Benchmark (2023, methodology): “Therefore, several tools for data auditing have been released recently: The ROOTS Search Tool (Piktus et al., 2023) and Data Por-traits (Marone and Durme, 2023) among others.”
“According to . . . ”: Prompting Language Models Improves Quoting from Pre-Training Data (2023, methodology): “We design a new metric called QUIP-Score, short for Qu oted I nformation P recision, which builds on D ATA P OR - TRAIT s and takes advantage of its speed and efficiency.”