RESEARCHInvestigateNEXT 12 MONTHS
Zero Gap Is Not Restoration: Stratified Per-Question Probability Evaluation and Step-wise Mitigation of Benchmark Contamination
arXiv cs.CL — Computation and Language
Factual evidence
What the source reports
Researchers expose flaws in standard LLM benchmark contamination metrics and propose a more accurate probability-based detection method.
Open source