REVIEW 5 cited by
Leak, Cheat, Repeat: Data Contamination and Evaluation Malpractices in Closed-Source LLMs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
abstract
Natural Language Processing (NLP) research is increasingly focusing on the use of Large Language Models (LLMs), with some of the most popular ones being either fully or partially closed-source. The lack of access to model details, especially regarding training data, has repeatedly raised concerns about data contamination among researchers. Several attempts have been made to address this issue, but they are limited to anecdotal evidence and trial and error. Additionally, they overlook the problem of \emph{indirect} data leaking, where models are iteratively improved by using data coming from users. In this work, we conduct the first systematic analysis of work using OpenAI's GPT-3.5 and GPT-4, the most prominently used LLMs today, in the context of data contamination. By analysing 255 papers and considering OpenAI's data usage policy, we extensively document the amount of data leaked to these models during the first year after the model's release. We report that these models have been globally exposed to $\sim$4.7M samples from 263 benchmarks. At the same time, we document a number of evaluation malpractices emerging in the reviewed papers, such as unfair or missing baseline comparisons and reproducibility issues. We release our results as a collaborative project on https://leak-llm.github.io/, where other researchers can contribute to our efforts.
Forward citations
Cited by 5 Pith papers
-
DecompSR: A dataset for decomposed analyses of compositional multihop spatial reasoning
DecompSR is a large, symbolically verified benchmark dataset and generation framework that independently varies productivity, substitutivity, overgeneralisation, and systematicity to probe compositional multihop spati...
-
AIMS.au: A Dataset for the Analysis of Modern Slavery Countermeasures in Corporate Statements
Introduces AIMS.au, a 5,731-statement, sentence-level annotated dataset for detecting disclosures mandated by Australia's Modern Slavery Act, with benchmarks showing fine-tuned models outperform zero-shot LLMs.
-
LessLeak-Bench: A First Investigation of Data Leakage in LLMs Across 83 Software Engineering Benchmarks
Across 83 SE benchmarks, average leakage into StarCoder's pretraining data is 4.8% (Python), 2.8% (Java), and 0.7% (C/C++), but QuixBugs and BigCloneBench are 100% and 55.7% leaked.
-
Reward Modeling for Reinforcement Learning-Based LLM Reasoning: Design, Challenges, and Evaluation
A taxonomy-driven survey arguing that reward design is the central mechanism shaping reliable LLM reasoning, with maps of reward paradigms, reward-hacking failure modes, and benchmark pitfalls.
-
Emergent LLM behaviors are observationally equivalent to data leakage
This paper shows that LLMs can verbally identify the coordination-game structure behind a published naming game and reproduce its likely outcome, arguing the observed 'emergent' conventions are observationally equival...
Discussion (0). Continue with ORCID to comment.