English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

When Are Likely Answers Right? On Sequence Probability and Correctness in LLMs

Forum topic · 小凯 · 2026-06-27

Summary

A new arXiv paper (2606.27359) by Johannes Zenn and Jonas Geiping investigates a fundamental question underlying LLM decoding methods: when does sequence probability—the conditional probability of a continuation given a prompt—actually align with correctness? Many decoding strategies work by shifting probability mass toward outputs the model considers more likely, either locally at the token level or globally at the sequence level. Their success therefore hinges on whether higher probability correlates with factual or task correctness. The authors set out to quantify this relationship across decoding methods and models, providing empirical analysis of when 'likely' answers are also 'right' answers. Posted on zhichai.net, the paper is relevant to researchers working on decoding strategies, sampling methods, and reliability of language model outputs.

Paper Overview

Research Area: Machine Learning

Authors: Johannes Zenn, Jonas Geiping

Published: 2026-06-27

arXiv: 2606.27359

Abstract

Many decoding methods for large language models can be understood as shifting probability mass toward outputs that are more likely under the model, either locally at the token level or globally at the sequence level. Therefore, their success depends on a fundamental question: when does sequence probability, that is, the conditional probability of a continuation given a prompt, actually align with correctness?

In this paper, the authors set out to quantify this relationship across decoding methods and models, empirically examining the conditions under which higher sequence probability corresponds to more correct model outputs. This analysis has direct implications for decoding strategies that exploit likelihood, from token-level methods to sequence-level approaches.

---

*Auto-collected on 2026-06-27*

Tags

#llm#decoding-methods#sequence-probability#arxiv#machine-learning#language-models

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208196