Saved in:
Bibliographic Details
Main Authors: Zhuang, Tianyi, Kuang, Chuqiao, Li, Xiaoguang, Teng, Yihua, Wu, Jihao, Wang, Yasheng, Shang, Lifeng
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.17807
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915171555868672
author Zhuang, Tianyi
Kuang, Chuqiao
Li, Xiaoguang
Teng, Yihua
Wu, Jihao
Wang, Yasheng
Shang, Lifeng
author_facet Zhuang, Tianyi
Kuang, Chuqiao
Li, Xiaoguang
Teng, Yihua
Wu, Jihao
Wang, Yasheng
Shang, Lifeng
contents We present DocPuzzle, a rigorously constructed benchmark for evaluating long-context reasoning capabilities in large language models (LLMs). This benchmark comprises 100 expert-level QA problems requiring multi-step reasoning over long real-world documents. To ensure the task quality and complexity, we implement a human-AI collaborative annotation-validation pipeline. DocPuzzle introduces an innovative evaluation framework that mitigates guessing bias through checklist-guided process analysis, establishing new standards for assessing reasoning capacities in LLMs. Our evaluation results show that: 1)Advanced slow-thinking reasoning models like o1-preview(69.7%) and DeepSeek-R1(66.3%) significantly outperform best general instruct models like Claude 3.5 Sonnet(57.7%); 2)Distilled reasoning models like DeepSeek-R1-Distill-Qwen-32B(41.3%) falls far behind the teacher model, suggesting challenges to maintain the generalization of reasoning capabilities relying solely on distillation.
format Preprint
id arxiv_https___arxiv_org_abs_2502_17807
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DocPuzzle: A Process-Aware Benchmark for Evaluating Realistic Long-Context Reasoning Capabilities
Zhuang, Tianyi
Kuang, Chuqiao
Li, Xiaoguang
Teng, Yihua
Wu, Jihao
Wang, Yasheng
Shang, Lifeng
Artificial Intelligence
We present DocPuzzle, a rigorously constructed benchmark for evaluating long-context reasoning capabilities in large language models (LLMs). This benchmark comprises 100 expert-level QA problems requiring multi-step reasoning over long real-world documents. To ensure the task quality and complexity, we implement a human-AI collaborative annotation-validation pipeline. DocPuzzle introduces an innovative evaluation framework that mitigates guessing bias through checklist-guided process analysis, establishing new standards for assessing reasoning capacities in LLMs. Our evaluation results show that: 1)Advanced slow-thinking reasoning models like o1-preview(69.7%) and DeepSeek-R1(66.3%) significantly outperform best general instruct models like Claude 3.5 Sonnet(57.7%); 2)Distilled reasoning models like DeepSeek-R1-Distill-Qwen-32B(41.3%) falls far behind the teacher model, suggesting challenges to maintain the generalization of reasoning capabilities relying solely on distillation.
title DocPuzzle: A Process-Aware Benchmark for Evaluating Realistic Long-Context Reasoning Capabilities
topic Artificial Intelligence
url https://arxiv.org/abs/2502.17807