Saved in:
Bibliographic Details
Main Authors: Zhang, Wenxuan, Jiang, Yuan-Hao, Cao, Yang, Wu, Yonghe
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.20356
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908801483931648
author Zhang, Wenxuan
Jiang, Yuan-Hao
Cao, Yang
Wu, Yonghe
author_facet Zhang, Wenxuan
Jiang, Yuan-Hao
Cao, Yang
Wu, Yonghe
contents Chunking strategies significantly impact the effectiveness of Retrieval-Augmented Generation (RAG) systems. Existing methods operate within fixed-granularity paradigms that rely on static boundary identification, limiting their adaptability to diverse query requirements. This paper presents FreeChunker, a Cross-Granularity Encoding Framework that fundamentally transforms the traditional chunking paradigm: the framework treats sentences as atomic units and shifts from static chunk segmentation to flexible retrieval supporting arbitrary sentence combinations. This paradigm shift not only significantly avoids the computational overhead required for semantic boundary detection, but also enhances adaptability to complex queries. Experimental evaluation on LongBench V2 demonstrates that FreeChunker possesses significant advantages in both retrieval performance and time efficiency compared to existing chunking methods. The pre-trained models and codes are available at https://github.com/mazehart/FreeChunker.
format Preprint
id arxiv_https___arxiv_org_abs_2510_20356
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FreeChunker: A Cross-Granularity Chunking Framework
Zhang, Wenxuan
Jiang, Yuan-Hao
Cao, Yang
Wu, Yonghe
Computation and Language
Chunking strategies significantly impact the effectiveness of Retrieval-Augmented Generation (RAG) systems. Existing methods operate within fixed-granularity paradigms that rely on static boundary identification, limiting their adaptability to diverse query requirements. This paper presents FreeChunker, a Cross-Granularity Encoding Framework that fundamentally transforms the traditional chunking paradigm: the framework treats sentences as atomic units and shifts from static chunk segmentation to flexible retrieval supporting arbitrary sentence combinations. This paradigm shift not only significantly avoids the computational overhead required for semantic boundary detection, but also enhances adaptability to complex queries. Experimental evaluation on LongBench V2 demonstrates that FreeChunker possesses significant advantages in both retrieval performance and time efficiency compared to existing chunking methods. The pre-trained models and codes are available at https://github.com/mazehart/FreeChunker.
title FreeChunker: A Cross-Granularity Chunking Framework
topic Computation and Language
url https://arxiv.org/abs/2510.20356