The paper introduces the ItD framework to improve the inductive capability of Large Language Models (LLMs) through deduction. It consists of two main components: Deductive Data Generation and Naive Bayesian Induction. The framework is tested on two types of induction tasks: Instruction Induction and List Function, showcasing significant performance improvements compared to existing methods. ItD effectively leverages the deductive capability of LLMs to enhance their inductive abilities.
Sang ngôn ngữ khác
từ nội dung nguồn
arxiv.org
Thông tin chi tiết chính được chắt lọc từ
by Wangtao Sun,... lúc arxiv.org 03-12-2024
https://arxiv.org/pdf/2403.05789.pdfYêu cầu sâu hơn