English

Batch Universal Prediction

Information Theory 2024-02-07 v1 Machine Learning math.IT Machine Learning

Abstract

Large language models (LLMs) have recently gained much popularity due to their surprising ability at generating human-like English sentences. LLMs are essentially predictors, estimating the probability of a sequence of words given the past. Therefore, it is natural to evaluate their performance from a universal prediction perspective. In order to do that fairly, we introduce the notion of batch regret as a modification of the classical average regret, and we study its asymptotical value for add-constant predictors, in the case of memoryless sources and first-order Markov sources.

Keywords

Cite

@article{arxiv.2402.03901,
  title  = {Batch Universal Prediction},
  author = {Marco Bondaschi and Michael Gastpar},
  journal= {arXiv preprint arXiv:2402.03901},
  year   = {2024}
}
R2 v1 2026-06-28T14:39:58.660Z