Stop Uploading Test Data in Plain Text: Practical Strategies for Mitigating Data Contamination by Evaluation Benchmarks

Alon Jacovi, Avi Caciularu, Omer Goldman, Yoav Goldberg

Submission Type: Regular Short Paper

Submission Track: Theme Track: Large Language Models and the Future of NLP

Submission Track 2: Resources and Evaluation

Keywords: data contamination, contamination, evaluation, test data, benchmarks, closed models, pretraining

TL;DR: Strategies (to be enacted by eval researchers) designed to prevent data contamination in the training of large language models

Abstract:

Data contamination has become prevalent and challenging with the rise of models pretrained on large automatically-crawled corpora.
For closed models, the training data becomes a trade secret, and even for open models, it is not trivial to detect contamination.
Strategies such as leaderboards with hidden answers, or using test data which is guaranteed to be unseen, are expensive and become fragile with time. Assuming that all relevant actors value clean test data and will cooperate to mitigate data contamination, what can be done?
We propose three strategies that can make a difference:
(1) Test data made public should be encrypted with a public key and licensed
to disallow derivative distribution;
(2) demand training exclusion controls from closed API holders, and protect your test data by refusing to evaluate without them;
(3) avoid data which appears with its solution on the internet, and release the web-page context of internet-derived data along with the data.
These strategies are practical and can be effective in preventing data contamination.

Submission Number: 1021