EvaluationAgents
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Google·January 28, 2022
Jason Wei, Xuezhi Wang, Dale Schuurmans
View on arXivTL;DR
Shows that prompting models to produce step-by-step reasoning dramatically improves performance on arithmetic, commonsense, and symbolic reasoning tasks.
Why it matters
Established reasoning-by-prompting as a core capability and seeded the line of work that became inference-time reasoning models.