Can I read Evaluating Reasoning Robustness in LLMs on EtoBox?
Evaluating Reasoning Robustness in LLMs by maryht1706 is a document available to read on EtoBox.
What is Evaluating Reasoning Robustness in LLMs about?
This document discusses the challenges of reasoning robustness in large language models (LLMs), highlighting significant performance degradation when faced with novel or incomplete data. It identifies four key limitations: positional bias, instruction sensitivity, numerical fragility, and memory dependence, which hinder systematic reasoning. To address these issues, the paper introduces a novel benchmark called Math-RoB, designed to assess and improve the reasoning capabilities of LLMs through diverse datas
- Author
- maryht1706
- Language
- EN