ACL Findings2026
Data Selection for Multi-turn Dialogue Instruction Tuning
Bo Li, Shikun Zhang, Wei Ye
Abstract
Instruction-tuned language models increasingly rely on large multi-turn dialogue corpora, but these datasets are often noisy and structurally inconsistent, with topic drift, repetitive chitchat, and mismatched answer formats across turns. We address this from a data selection perspective and propose MDS (Multi-turn Dialogue Selection), a dialogue-level framework that scores whole conversations rather than isolated turns. MDS combines a global coverage stage that performs bin-wise selection in the user-query trajectory space to retain representative yet nonredundant dialogues, with a local structural stage that evaluates within-dialogue reliability through entity-grounded topic grounding and information progress, together with queryanswer form consistency for functional alignment. MDS outperforms strong single-turn selectors, dialogue-level LLM scorers, and heuristic baselines on three multi-turn benchmarks and an in-domain Banking test set, achieving the best overall rank across reference-free and reference-based metrics, and is more robust on long conversations under the same training budget. Code and resources are included in the supplementary materials.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
3 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Chain of Evidence: Pixel-Level Visual Attribution for Iterative Retrieval-Augmented Generation cites Instruction Data Selection via Answer Divergence
- Instruction Data Selection via Answer Divergence cites LESS: Selecting Influential Data for Targeted Instruction Tuning
- Beyond Semantic Relevance: Counterfactual Risk Minimization for Robust Retrieval-Augmented Generation cites Instruction Data Selection via Answer Divergence
Cited by top-tier papers3
- Instruction Data Selection via Answer DivergenceBo Li, Mingda Wang, Shikun Zhang, Wei YeACL 2026 · 4 citations
- Chain of Evidence: Pixel-Level Visual Attribution for Iterative Retrieval-Augmented GenerationPeiyang Liu, Ziqiang Cui, Xi Wang, Di Liang et al.SIGIR 2026
- Beyond Semantic Relevance: Counterfactual Risk Minimization for Robust Retrieval-Augmented GenerationPeiyang Liu, Qiang Yan, Ziqiang Cui, Di Liang et al.SIGIR 2026
Builds on21
- LIMA: Less Is More for AlignmentChunting Zhou, Pengfei Liu, Puxin Xu, Srinivasan Iyer et al.NeurIPS 2023 · 1,486 citations
- Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!Xiangyu Qi, Yi Zeng, Tinghao Xie, Pin-Yu Chen et al.ICLR 2024 · 1,104 citations
- LESS: Selecting Influential Data for Targeted Instruction TuningMengzhou Xia, Sadhika Malladi, Suchin Gururangan, Sanjeev Arora et al.ICML 2024 · 460 citations
- LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation DatasetLianmin Zheng, Wei-Lin Chiang, Ying Sheng, Tianle Li et al.ICLR 2024 · 419 citations
- OpenChat: Advancing Open-source Language Models with Mixed-Quality DataGuan Wang, Sijie Cheng, Xianyuan Zhan, Xiangang Li et al.ICLR 2024 · 328 citations
Related papers
- Evaluating Language Models in Realistic Conversational ContextsIlija Subasic, Andrew Rabinovich, Zhao ChenICML 2026
- MARS-Bench: A Multi-turn Athletic Real-world Scenario Benchmark for Dialogue EvaluationChenghao Yang, Yinbo Luo, Zhoufutu Wen, Qi Chu et al.EMNLP Findings 2025 · 1 citation
- MT-Bench-101: A Fine-Grained Benchmark for Evaluating Large Language Models in Multi-Turn DialoguesGe Bai, Jie Liu, Xingyuan Bu, Yancheng He et al.ACL 2024 · 35 citations
- Review-Instruct: A Review-Driven Multi-Turn Conversations Generation Method for Large Language ModelsJiangxu Wu, Cong Wang, Tianhuang Su, Jun Yang et al.ACL Findings 2025
- ConsistentChat: Building Skeleton-Guided Consistent Multi-Turn Dialogues for Large Language Models from ScratchJiawei Chen, Xinyan Guan, Qianhao Yuan, Guozhao Mo et al.EMNLP 2025
