ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2305.02633
67
20

Conformal Nucleus Sampling

4 May 2023
Shauli Ravfogel
Carlos Wert Carvajal
M.F. Eggl
    UQLM
ArXivPDFHTML
Abstract

Language models generate text based on successively sampling the next word. A decoding procedure based on nucleus (top-ppp) sampling chooses from the smallest possible set of words whose cumulative probability exceeds the probability ppp. In this work, we assess whether a top-ppp set is indeed aligned with its probabilistic meaning in various linguistic contexts. We employ conformal prediction, a calibration procedure that focuses on the construction of minimal prediction sets according to a desired confidence level, to calibrate the parameter ppp as a function of the entropy of the next word distribution. We find that OPT models are overconfident, and that calibration shows a moderate inverse scaling with model size.

View on arXiv
Comments on this paper