Long-Document QA with Chain-of-Structured-Thought and Fine-Tuned SLMs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liang, Zhuowen, Lin, Xiaotian, Zhang, Zhengxuan, Luo, Yuyu, Wang, Haixun, Tang, Nan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917372494872576
author Liang, Zhuowen
Lin, Xiaotian
Zhang, Zhengxuan
Luo, Yuyu
Wang, Haixun
Tang, Nan
author_facet Liang, Zhuowen
Lin, Xiaotian
Zhang, Zhengxuan
Luo, Yuyu
Wang, Haixun
Tang, Nan
contents Large language models (LLMs) are widely applied to data analytics over documents, yet direct reasoning over long, noisy documents remains brittle and error-prone. Hence, we study document question answering (QA) that consolidates dispersed evidence into a structured output (e.g., a table, graph, or chunks) to support reliable, verifiable QA. We propose a two-pillar framework, LiteCoST, to achieve both high accuracy and low latency with small language models (SLMs). Pillar 1: Chain-of-Structured-Thought (CoST). We introduce a CoST template, a schema-aware instruction that guides a strong LLM to produce both a step-wise CoST trace and the corresponding structured output. The process induces a minimal structure, normalizes entities/units, aligns records, serializes the output, and verifies/refines it, yielding auditable supervision. Pillar 2: SLM fine-tuning. The compact models are trained on LLM-generated CoST data in two stages: Supervised Fine-Tuning for structural alignment, followed by Group Relative Policy Optimization (GRPO) incorporating triple rewards for answer/format quality and process consistency. By distilling structure-first behavior into SLMs, this approach achieves LLM-comparable quality on multi-domain long-document QA using 3B/7B SLMs, while delivering 2-4x lower latency than GPT-4o and DeepSeek-R1 (671B). The code is available at https://github.com/HKUSTDial/LiteCoST.
format Preprint
id arxiv_https___arxiv_org_abs_2603_29232
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Long-Document QA with Chain-of-Structured-Thought and Fine-Tuned SLMs
Liang, Zhuowen
Lin, Xiaotian
Zhang, Zhengxuan
Luo, Yuyu
Wang, Haixun
Tang, Nan
Computation and Language
Artificial Intelligence
Machine Learning
Large language models (LLMs) are widely applied to data analytics over documents, yet direct reasoning over long, noisy documents remains brittle and error-prone. Hence, we study document question answering (QA) that consolidates dispersed evidence into a structured output (e.g., a table, graph, or chunks) to support reliable, verifiable QA. We propose a two-pillar framework, LiteCoST, to achieve both high accuracy and low latency with small language models (SLMs). Pillar 1: Chain-of-Structured-Thought (CoST). We introduce a CoST template, a schema-aware instruction that guides a strong LLM to produce both a step-wise CoST trace and the corresponding structured output. The process induces a minimal structure, normalizes entities/units, aligns records, serializes the output, and verifies/refines it, yielding auditable supervision. Pillar 2: SLM fine-tuning. The compact models are trained on LLM-generated CoST data in two stages: Supervised Fine-Tuning for structural alignment, followed by Group Relative Policy Optimization (GRPO) incorporating triple rewards for answer/format quality and process consistency. By distilling structure-first behavior into SLMs, this approach achieves LLM-comparable quality on multi-domain long-document QA using 3B/7B SLMs, while delivering 2-4x lower latency than GPT-4o and DeepSeek-R1 (671B). The code is available at https://github.com/HKUSTDial/LiteCoST.
title Long-Document QA with Chain-of-Structured-Thought and Fine-Tuned SLMs
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2603.29232