ReSURE: Regularizing Supervision Unreliability for Multi-turn Dialogue Fine-tuning (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods address data quality via static prefiltering, which decouples quality control from training and fails to mitigate turn-level error propagation. |
| Approach: | They propose an adaptive learning method that dynamically down-weights unreliable supervision without explicit filtering. |
| Outcome: | Experiments on single-source and mixed-quality datasets show improved stability and response quality. |
Similar Papers
Correctable-DST: Mitigating Historical Context Mismatch between Training and Inference for Improved Dialogue State Tracking (2022.emnlp-main)
Copied to clipboard
Hongyan Xie, Haoxiang Su, Shuangyong Song, Hao Huang, Bo Zou, Kun Deng, Jianghua Lin, Zhihui Zhang, Xiaodong He
| Challenge: | Existing dialogue state tracking approaches predict the dialogue state of a target turn sequentially based on the ground-truth previous dialogue state. |
| Approach: | They propose a method that predicts dialogue state sequentially based on previous dialogue state . they propose generating a previously “predicted” dialogue state using ground-truth previous dialogue states . |
| Outcome: | The proposed method achieves 67.51%, 68.24%, 70.30%, 71.38%, and 81.27% joint goal accuracy on MultiWOZ 2.0-2.4 datasets. |
How to Mitigate Overfitting in Weak-to-strong Generalization? (2025.acl-long)
Copied to clipboard
| Challenge: | Experimental results show that weak-to-strong generalization significantly improves PGR compared to naive weak- to-strong . superalignment refers to how humans can align models on tasks beyond human ability to evaluate . |
| Approach: | They propose a framework that elicits the capabilities of strong models through weak supervisors . they propose 'superalignment' to ensure that strong models align with supervisors' intentions . |
| Outcome: | The proposed framework significantly improves quality of supervision signals and quality of input questions compared to naive weak-to-strong generalization . |
Improving Dialogue State Tracking with Turn-based Loss Function and Sequential Data Augmentation (2021.findings-emnlp)
Copied to clipboard
| Challenge: | Existing models rely on a traditional cross-entropy loss function during training, which may not be optimal for improving the joint goal accuracy. |
| Approach: | They propose a Turn-based Loss Function that penalises the model if it inaccurately predicts a slot value at the early turns more so than in later turns to improve joint goal accuracy. |
| Outcome: | The proposed techniques improve the state-of-the-art model by approximately 7-8% relative reduction in error and achieve a new state- of-the art joint goal accuracy with 59.50 and 54.90 on MultiWOZ2.1 and MultiWOz2.2, respectively. |
Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards (2026.acl-long)
Copied to clipboard
Ming Li, Pei Chen, Zhenhao Zhang, Tao Yang, Xinyang Zhang, Han Li, Tianyu Cao, Ming Zeng, Zhuofeng Wu, Meng Jiang, Huasheng Li, Lihong Li, Bing Yin
| Challenge: | Large Language Models exhibit strong capabilities in single-turn instruction following but suffer from Lost-in-Conversation (LiC) when instructions are revealed progressively in multi-turn settings, models get "Lost in Conversation" |
| Approach: | They propose a framework that encourages models to generate correct answers and judge solvability in multi-turn conversations. |
| Outcome: | The proposed framework improves models' ability to balance problem-solving with abstention . it reduces premature answering behaviors that cause lost-in-conversation (LiC) |
Confidence Should Be Calibrated More Than One Turn Deep (2026.acl-long)
Copied to clipboard
| Challenge: | Existing work on confidence estimation and calibration focuses on single-turn settings . existing work on multi-turn calibration ignores the risks and potential of multi-turned conversations . |
| Approach: | They propose a multi-turn calibration task that reframes calibration from a static property into a dynamic challenge central to reliable multi- turn conversations. |
| Outcome: | The proposed model minimizes ECE@T and leverages ConfChat to improve confidence . the proposed model preserves and even enhances model performance in multi-turn interactions. |
Data Selection for Multi-turn Dialogue Instruction Tuning (2026.findings-acl)
Copied to clipboard
| Challenge: | Instruction-tuned language models often use noisy multi-turn dialogue datasets with topic drift, repetitive chitchat, and mismatched answer formats across turns. |
| Approach: | They propose a dialogue-level framework that scores whole conversations rather than isolated turns. |
| Outcome: | The proposed framework outperforms strong single-turn selectors, dialogue-level LLM scorers and heuristic baselines on three multi-turn benchmarks and an in-domain Banking test set. |
One Battle After Another: Probing LLMs’ Limits on Multi-Turn Instruction Following with a Benchmark Evolving Framework (2026.acl-long)
Copied to clipboard
| Challenge: | Existing benchmarks for instruction-following in multi-topic dialogues are limited to a fixed number of turns, susceptible to saturation and failing to account for users’ interactive experience. |
| Approach: | They propose a framework featuring a three-layer tracking mechanism and a query synthesis agent to mimic sequential user behaviors. |
| Outcome: | The proposed framework outperforms existing benchmarks in the evaluation of instruction following in multi-topic dialogues and demonstrates deficiencies in failure recovery and fine-grained instruction following. |
Assertion-Conditioned Compliance: A Provenance-Aware Vulnerability in Multi-Turn Tool-Calling Agents (2026.eacl-industry)
Copied to clipboard
| Challenge: | Multi-turn tool-calling models have emerged as a key feature in modern AI assistants, but their success in safety-critical industries remains constrained by concerns about transparency and model resilience. |
| Approach: | They propose a new evaluation paradigm for multi-turn function-calling LLMs that provides holistic metrics that evaluate a model’s behavior when confronted with misleading assertions. |
| Outcome: | The proposed evaluation paradigm evaluates a model's behavior when confronted with misleading assertions originating from two distinct vectors: (1) user-sourced assertions (USAs) and (2) function-sourced assertions (FSAs). |
Generalizing Trust: Weak-to-Strong Trustworthiness in Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | Recent studies have highlighted weak-to-strong generalization, where a strong model trained only on a weak model’s labels surpasses the weak model in task performance. |
| Approach: | They propose two fundamental fine-tuning strategies that leverage trustworthiness regularization during the fine-uning of the weak model and the weak-to-strong transfer to improve trustworthy. |
| Outcome: | The proposed models show that they can generalize robustness, fairness, and privacy better when trained on weak models than models trained on strong models. |
Review-Instruct: A Review-Driven Multi-Turn Conversations Generation Method for Large Language Models (2025.findings-acl)
Copied to clipboard
Jiangxu Wu, Cong Wang, Tianhuang Su, Lin Haozhi, JunYang JunYang, Zhangchao Zhangchao, Binqiang Pan, SongpanYang SongpanYang, Mingpeng Mingpeng, Kai Shi, Zixian Li
| Challenge: | Existing methods for generating multi-turn dialogue data struggle to ensure both diversity and quality in instructions. |
| Approach: | They propose a framework that synthesizes multi-turn conversations through an iterative "Ask-Respond-Review" process involving three agent roles: a Candidate, multiple Reviewers, and a Chairman. |
| Outcome: | The proposed framework synthesizes multi-turn conversations through an iterative "Ask-Respond-Review" process involving three agent roles: a Candidate, multiple Reviewers, and a Chairman. |