Testing for LLM response differences: the case of a composite null consisting of semantically irrelevant query perturbations

Documenti analoghi

• When prompt perturbations break your A/B test: A valid statistical test for generative surveying
  di: Helm, Hayden, et al.
  Pubblicazione: (2026)
• Consistent estimation of generative model representations in the data kernel perspective space
  di: Acharyya, Aranyak, et al.
  Pubblicazione: (2024)
• Query-efficient model evaluation using cached responses
  di: Helm, Hayden, et al.
  Pubblicazione: (2026)
• Comparing Foundation Models using Data Kernels
  di: Duderstadt, Brandon, et al.
  Pubblicazione: (2023)
• Convergence guarantees for response prediction for latent structure network time series
  di: Acharyya, Aranyak, et al.
  Pubblicazione: (2025)