Mind the Quote: Enabling Quotation-Aware Dialogue in LLMs via Plug-and-Play Modules

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Yueqi, Yuan, Peiwen, Feng, Shaoxiong, Li, Yiwei, Wang, Xinglin, Shi, Jiayi, Tan, Chuyi, Pan, Boyuan, Hu, Yao, Li, Kan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910975219728384
author Zhang, Yueqi
Yuan, Peiwen
Feng, Shaoxiong
Li, Yiwei
Wang, Xinglin
Shi, Jiayi
Tan, Chuyi
Pan, Boyuan
Hu, Yao
Li, Kan
author_facet Zhang, Yueqi
Yuan, Peiwen
Feng, Shaoxiong
Li, Yiwei
Wang, Xinglin
Shi, Jiayi
Tan, Chuyi
Pan, Boyuan
Hu, Yao
Li, Kan
contents Human-AI conversation frequently relies on quoting earlier text-"check it with the formula I just highlighted"-yet today's large language models (LLMs) lack an explicit mechanism for locating and exploiting such spans. We formalise the challenge as span-conditioned generation, decomposing each turn into the dialogue history, a set of token-offset quotation spans, and an intent utterance. Building on this abstraction, we introduce a quotation-centric data pipeline that automatically synthesises task-specific dialogues, verifies answer correctness through multi-stage consistency checks, and yields both a heterogeneous training corpus and the first benchmark covering five representative scenarios. To meet the benchmark's zero-overhead and parameter-efficiency requirements, we propose QuAda, a lightweight training-based method that attaches two bottleneck projections to every attention head, dynamically amplifying or suppressing attention to quoted spans at inference time while leaving the prompt unchanged and updating < 2.8% of backbone weights. Experiments across models show that QuAda is suitable for all scenarios and generalises to unseen topics, offering an effective, plug-and-play solution for quotation-aware dialogue.
format Preprint
id arxiv_https___arxiv_org_abs_2505_24292
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mind the Quote: Enabling Quotation-Aware Dialogue in LLMs via Plug-and-Play Modules
Zhang, Yueqi
Yuan, Peiwen
Feng, Shaoxiong
Li, Yiwei
Wang, Xinglin
Shi, Jiayi
Tan, Chuyi
Pan, Boyuan
Hu, Yao
Li, Kan
Artificial Intelligence
Computation and Language
Human-AI conversation frequently relies on quoting earlier text-"check it with the formula I just highlighted"-yet today's large language models (LLMs) lack an explicit mechanism for locating and exploiting such spans. We formalise the challenge as span-conditioned generation, decomposing each turn into the dialogue history, a set of token-offset quotation spans, and an intent utterance. Building on this abstraction, we introduce a quotation-centric data pipeline that automatically synthesises task-specific dialogues, verifies answer correctness through multi-stage consistency checks, and yields both a heterogeneous training corpus and the first benchmark covering five representative scenarios. To meet the benchmark's zero-overhead and parameter-efficiency requirements, we propose QuAda, a lightweight training-based method that attaches two bottleneck projections to every attention head, dynamically amplifying or suppressing attention to quoted spans at inference time while leaving the prompt unchanged and updating < 2.8% of backbone weights. Experiments across models show that QuAda is suitable for all scenarios and generalises to unseen topics, offering an effective, plug-and-play solution for quotation-aware dialogue.
title Mind the Quote: Enabling Quotation-Aware Dialogue in LLMs via Plug-and-Play Modules
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2505.24292