paper-with-me

홈 › Papers

PromptTailor: Multi-turn Intent-Aligned Prompt Synthesis for Lightweight LLMs

2025-11-20 · Yizhou Xu, Janet Davis arxiv

Lightweight language models remain attractive for on-device and privacy-sensitive applications, but their responses are highly sensitive to prompt quality. For open-ended generation, non-expert users often lack the knowledge or time to consistently craft high-quality prompts, leading them to rely on prompt optimization tools. However, a key challenge is ensuring the optimized prompts genuinely align with users' original intents and preferences. We introduce PromptTailor, a system for controllable prompt generation for open-ended text that improves model output quality by intent-aligned prompt synthesis. PromptTailor expands minimal user instructions into rich, domain-aware prompts while preserving the user's stated preferences. The system is a quantized Llama3-8B model fine-tuned with a lightweight LoRA adapter on 12,300 prompt-refinement dialogues spanning 41 everyday domains, distilled from three stronger LLMs. The adapter attaches to any Llama3-8B base, enabling edge deployment. In human and LLM-judge evaluations across multiple target models and optimization baselines, PromptTailor yields higher preference rates than chain-of-thought prompting and matches or surpasses state-of-the-art prompt optimization methods while requiring fewer model calls (e.g., 3 vs. 9). These results show that a compact student, guided by powerful teachers, can learn effective prompt-generation strategies that enhance response quality while maintaining alignment with user intent.

📄 PDF Abstract BibTeX arXiv:2511.21725

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

SEMA: Simple yet Effective Learning for Multi-Turn Jailbreak Attacks

2026-02-06 · Mingqian Feng, Xiaodong Liu, Weiwei Yang, Jialin Song 외 arxiv

Multi-turn jailbreaks capture the real threat model for safety-aligned chatbots, where single-turn attacks are merely a special case. Yet existing approaches break under exploration complexity and intent drift. We propos…

Reinforcement Learning

MultiBreak: A Scalable and Diverse Multi-turn Jailbreak Benchmark for Evaluating LLM Safety

2026-05-03 · Jialin Song, Xiaodong Liu, Weiwei Yang, Wuyang Chen 외 arxiv

We present MultiBreak, a scalable and diverse multi-turn jailbreak benchmark to evaluate large language model (LLM) safety. Multi-turn jailbreaks mimic natural conversational settings, making them easier to bypass safety…

Active Learning

Derail Yourself: Multi-turn LLM Jailbreak Attack through Self-discovered Clues

2024-10-14 · Qibing Ren, Hao Li, Dongrui Liu, Zhanxu Xie 외

This study exposes the safety vulnerabilities of Large Language Models (LLMs) in multi-turn interactions, where malicious users can obscure harmful intents across several queries. We introduce ActorAttack, a novel multi-…

LLM JailbreakSafety Alignment

SAMPO-Path: Segmentation Intent-Aligned Preference Optimization for Pathology Foundation Model Segmentation

2025-08-04 · Yonghuang Wu, Wenwen Zeng, Xuan Xie, Chengqian Zhao 외 arxiv

Foundation models have shown strong performance in multi-object segmentation with visual prompts, yet histopathology images remain challenging due to high cellular density, heterogeneity, and the gap between pixel-level …

Object Segmentation

Proactive Dialogue Model with Intent Prediction

2026-04-30 · Yang Luo arxiv

Dialogue models are inherently reactive, responding to the current user turn without anticipating upcoming intents, which leads to redundant interactions in multi-intent settings. We address this limitation by introducin…