How Contaminated Is Your Benchmark? Quantifying Dataset Leakage in Large Language Models with Kernel Divergence

Documenti analoghi

• ARGS: Alignment as Reward-Guided Search
  di: Khanov, Maxim, et al.
  Pubblicazione: (2024)
• PICLe: Eliciting Diverse Behaviors from Large Language Models with Persona In-Context Learning
  di: Choi, Hyeong Kyu, et al.
  Pubblicazione: (2024)
• Benchmarking Benchmark Leakage in Large Language Models
  di: Xu, Ruijie, et al.
  Pubblicazione: (2024)
• Safety-Aware Fine-Tuning of Large Language Models
  di: Choi, Hyeong Kyu, et al.
  Pubblicazione: (2024)
• Preference Leakage: A Contamination Problem in LLM-as-a-judge
  di: Li, Dawei, et al.
  Pubblicazione: (2025)