Authors:
Yuri Bezerra
and
Li Weigang
Affiliation:
TransLab, Department of Computer Science, University of Brasilia, Brasilia, Federal District, Brazil
Keyword(s):
Knowledge Distillation, Large Language Models, LLM Reasoning, Low-Rank Adaptation, Retrieval-Augmented Generation.
Abstract:
We introduce LLMQuoter, a lightweight, distillation-based model designed to enhance Retrieval-Augmented Generation (RAG) by extracting the most relevant textual evidence for downstream reasoning tasks. Built on the LLaMA-3B architecture and fine-tuned with Low-Rank Adaptation (LoRA) on a 15,000-sample subset of HotpotQA, LLMQuoter adopts a “quote-first-then-answer” strategy, efficiently identifying key quotes before passing curated snippets to reasoning models. This workflow reduces cognitive overhead and outperforms full-context approaches like Retrieval-Augmented Fine-Tuning (RAFT), achieving over 20-point accuracy gains across both small and large language models. By leveraging knowledge distillation from a high-performing teacher model, LLMQuoter achieves competitive results in a resource-efficient fine-tuning setup. It democratizes advanced RAG capabilities, delivering significant performance improvements without requiring extensive model retraining. Our results highlight the po
tential of distilled quote-based reasoning to streamline complex workflows, offering a scalable and practical solution for researchers and practitioners alike.
(More)