![[2504.04717] Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Models](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2504.04717] Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Models
Summary
This article surveys advancements in multi-turn interactions with large language models (LLMs), focusing on evaluation methods, challenges, and future research directions.
Why It Matters
As applications of LLMs evolve, understanding multi-turn interactions becomes crucial for improving user experience and effectiveness in real-world scenarios. This survey highlights the current state of research, identifies challenges, and suggests paths for future development, making it a valuable resource for researchers and practitioners in AI and NLP.
Key Takeaways
- Multi-turn interactions are essential for real-world applications of LLMs.
- The survey categorizes existing benchmarks and datasets for evaluating multi-turn dialogues.
- It reviews various enhancement methodologies, including model-centric and external integration approaches.
- Identifies challenges in maintaining context, coherence, and responsiveness in dialogues.
- Suggests future research directions to improve multi-turn interaction robustness.
Computer Science > Computation and Language arXiv:2504.04717 (cs) [Submitted on 7 Apr 2025 (v1), last revised 21 Feb 2026 (this version, v5)] Title:Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Models Authors:Yubo Li, Xiaobin Shen, Xinyu Yao, Xueying Ding, Yidi Miao, Ramayya Krishnan, Rema Padman View a PDF of the paper titled Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Models, by Yubo Li and 6 other authors View PDF HTML (experimental) Abstract:Recent advances in large language models (LLMs) have substantially improved single-turn task performance, yet real-world applications increasingly demand sophisticated multi-turn interactions. This survey provides a comprehensive review of recent progress in evaluating and enhancing multi-turn LLM interactions. Centered on a task-oriented taxonomy-spanning instruction following in domains such as mathematics and coding, and conversational engagement in role-playing, healthcare, education, and adversarial jailbreak settings-we systematically examine the challenges of maintaining context, coherence, fairness, and responsiveness across prolonged dialogues. We organize existing benchmarks and datasets into coherent categories reflecting the evolving landscape of multi-turn dialogue evaluation, and review a broad spectrum of enhancement methodologies, including model-centric strategies (in-context learning, supervised fine-tuning, reinforcement learning, and architectural inn...
