Learn Before
Multiple Choice

A language model is tasked with generating a step-by-step solution to a logic puzzle. The puzzle has several plausible initial moves, but an incorrect choice early on will lead to an unsolvable state later. A simple decoding approach that always commits to the single most probable next step often fails on this task. Which of the following alternative strategies is best designed to overcome this specific challenge?

0

1

Updated 2025-10-02

Contributors are:

Who are from:

Tags

Ch.5 Inference - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Analysis in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science