Context-CoT: Enhancing Context Learning via High-Quality Reasoning Synthesis
人工智能
2026-05-26 v1
摘要
While LLMs excel at reasoning over prompts using static pretrained knowledge, they struggle significantly with context learning-the ability to dynamically extract, internalize, and apply new knowledge from complex, task-specific contexts. Recent evaluations on the CL-Bench reveal a critical capability gap: frontier models solve only 17.2% of context-dependent tasks on average.
引用
@article{arxiv.2605.25354,
title = {Context-CoT: Enhancing Context Learning via High-Quality Reasoning Synthesis},
author = {Hongbo Jin and Mingnan Zhu and Jingqi Tian and Xu Jiang and Zhongjing Du and Haoran Tang and Siyi Xie and Qiaoman Zhang and Jiayu Ding},
journal= {arXiv preprint arXiv:2605.25354},
year = {2026}
}