Large Language Models (LLMs) have shown remarkable performance in multi-turn dialogue. However, in multi-turn dialogue, models still struggle to stay aligned with what has been established earlier, follow dependencies across many turns, and avoid drifting into incorrect facts as the interaction grows longer. Existing approaches primarily focus on extending the context window, introducing external memory, or applying context compression, yet these methods still face limitations such as contextual inertia and state drift. To address these challenges, we propose the Adaptive Context Refactoring (ACR) Framework, which dynamically monitors and reshapes the interaction history to mitigate contextual inertia and state drift actively. ACR is built on a library of context refactoring operators and a teacher-guided self-evolving training paradigm that learns when to intervene and how to refactor, thereby decoupling context management from the reasoning process. Extensive experiments on multi-turn dialogue demonstrate that our method significantly outperforms existing baselines while reducing token consumption. Our code is available at https://github.com/ClannadKno/multi-turn.
论文检索
输入标题、作者或关键词，从 1,444 篇学术成果中精准定位
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 non-redundant dialogues, with a local structural stage that evaluates within-dialogue reliability through entity-grounded topic grounding and information progress, together with query-answer 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.
Existing benchmarks treat multi-turn conversation and reasoning-intensive retrieval separately, yet real-world information seeking requires both. To bridge this gap, we present a benchmark for reasoning-based conversational information retrieval comprising 707 conversations (2,971 turns) across eleven domains. To ensure quality, our Decomposition-and-Verification framework transforms complex queries into fact-grounded multi-turn dialogues through multi-level validation, where atomic facts are verified against sources and explicit retrieval reasoning is generated for each turn. Comprehensive evaluation reveals that combining conversation history with reasoning doubles retrieval performance (Baseline .236 \rightarrow History+Reasoning .479 nDCG@10), while reasoning-specialized models substantially outperform dense encoders. Despite these gains, further analysis highlights that implicit reasoning remains challenging, particularly when logical connections are not explicitly stated in the text. [<https://github.com/RECOR-Benchmark/RECOR>]
WavAlign: Enhancing Intelligence and Expressiveness in Spoken Dialogue Models via Adaptive Hybrid Post-Training
PDF ↗End-to-end spoken dialogue models have garnered significant attention because they offer a higher potential ceiling in expressiveness and perceptual ability than cascaded systems. However, the intelligence and expressiveness of current open-source spoken dialogue models often remain below expectations. Motivated by the success of online reinforcement learning(RL) in other domains, one might attempt to directly apply preference optimization to spoken dialogue models, yet this transfer is non-trivial. We analyze these obstacles from the perspectives of reward modeling and rollout sampling, focusing on how sparse preference supervision interacts with dense speech generation under shared-parameter updates. Based on the analysis, we propose a modality-aware adaptive post-training recipe that makes RL practical for spoken dialogue: it constrains preference updates to the semantic channel and improves acoustic behavior via explicit anchoring, while dynamically regulating their mixture from rollout statistics to avoid unreliable preference gradients. We evaluate the method across multiple spoken dialogue benchmarks and representative architectures, and observe consistent improvements in semantic quality and speech expressiveness.
MTMCS-Bench: Evaluating Contextual Safety of Multimodal Large Language Models in Multi-Turn Dialogues
PDF ↗Multimodal large language models (MLLMs) are increasingly deployed as assistants that interact through text and images, making it crucial to evaluate contextual safety when risk depends on both the visual scene and the evolving dialogue. Existing contextual safety benchmarks are mostly single-turn and often miss how malicious intent can emerge gradually or how the same scene can support both benign and exploitative goals. We introduce the Multi-Turn Multimodal Contextual Safety Benchmark (MTMCS-Bench), a benchmark of realistic images and multi-turn conversations that evaluates contextual safety in MLLMs under two complementary settings, escalation-based risk and context-switch risk. MTMCS-Bench offers paired safe and unsafe dialogues with structured evaluation. It contains over 30 thousand multimodal (image+text) and unimodal (text-only) samples, with metrics that separately measure contextual intent recognition, safety-awareness on unsafe cases, and helpfulness on benign ones. Across eight open-source and seven proprietary MLLMs, we observe persistent trade-offs between contextual safety and utility, with models tending to either miss gradual risks or over-refuse benign dialogues. Finally, we evaluate five current guardrails and find that they mitigate some failures but do not fully resolve multi-turn contextual risks.
Role-playing models (RPMs) are widely used in real-world applications but underperform when deployed in the wild. This degradation can be attributed to distribution shifts, including user, character, and dialogue compositional shifts. Existing methods like LLM-as-a-judge fall short in providing a fine-grained diagnosis of how these shifts affect RPM generalization, and thus there lack formal frameworks to characterize RPM generalization behaviors. To bridge these gaps, we introduce an information-theoretic metric, named reasoning-based effective mutual information difference (R-EMID), to measure RPM performance degradation in an interpretable way. We also derive an upper bound on R-EMID to predict the worst-case generalization performance of RPMs and theoretically reveal how various shifts contribute to the RPM performance degradation. Moreover, we propose a co-evolving reinforcement learning framework to adaptively model the connection among user, character, and dialogue context and thus enhance the estimation of dialogue response generation probability, which is critical for calculating R-EMID. Finally, we evaluate the generalization performance of various RPMs using R-EMID, finding that user shift poses the highest risk among all shifts and reinforcement learning is the most effective approach for enhancing RPM generalization. Code and data are available at https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/RPM-Generalization.
Large language models (LLMs) increasingly rely on external tools to complete complex tasks, yet their ability to recognize and correct their own tool-use mistakes remains underexplored. Existing benchmarks primarily evaluate planning and execution success, overlooking the self-reflective dimension of tool use. To address this gap, we present ReflecTool-Bench, the first benchmark designed to assess LLMs’ self-reflective reasoning in tool-augmented multi-turn dialogues. ReflecTool-Bench covers 10 domains with 88 distinct APIs and 968 annotated dialogues, systematically injecting diverse error types arising from both user and assistant behavior. The benchmark defines two complementary evaluation setups: the Critique task, where models diagnose errors in third-party dialogues, and the Self-Reflection Task, where models must detect and repair their own prior tool-use mistakes. We introduce fine-grained metrics for error detection, error classification, correction accuracy, and explanation quality, enabling a holistic assessment of reflective reasoning. Evaluations across 12 state-of-the-art models, including both API-based closed source models and open source models, reveal that while models can reliably identify user-originated errors, they struggle with assistant-originated ones, and performance drops sharply when moving from critique to self-reflection.
The capabilities of Large Language Models (LLMs) are limited to some extent by pre-training, so some researchers optimize LLMs through post-training. Existing post-training strategies, such as memory-based retrieval or preference optimization, improve user alignment yet fail to enhance the model’s domain cognition. To bridge this gap, we propose a novel Dual-Phase Self-Evolution (DPSE) framework that jointly optimizes user preference adaptation and domain-specific competence. DPSE introduces a Censor module to extract multi-dimensional interaction signals and estimate satisfaction scores, which guide structured data expansion via topic-aware and preference-driven strategies. These expanded datasets support a two-stage fine-tuning pipeline: supervised domain grounding followed by frequency-aware preference optimization. Experiments across general NLP benchmarks and long-term dialogue tasks demonstrate that DPSE consistently outperforms Supervised Fine-Tuning, Preference Optimization, and Memory-Augmented baselines. Ablation studies validate the contribution of each module. In this way, our framework provides an autonomous path toward continual self-evolution of LLMs.
MTO: A Multi-turn Conversational Text-to-OverpassQL Dataset for Enhanced OpenStreetMap Query Generation
PDF ↗We propose a comprehensive framework for constructing multi-turn Text-to-OverpassQL dialogue datasets. Under this framework, we introduce the first multi-turn Text-to-OverpassQL dataset built upon the OverpassNL corpus. Our dataset comprises over 7,800 dialogues, each containing 2 to 4 user utterances, resulting in more than 20,000 individual utterances aligned with executable Overpass queries. To generate high-quality multi-turn dialogues, we design a four-stage pipeline. First, we convert Overpass queries into syntax trees using a custom parser developed based on the official OverpassQL grammar. This enables structural manipulation while preserving syntactic and executable validity. Second, we apply a diverse set of tree-editing templates, including both simple keyword-level changes and complex structural decompositions, to produce multiple valid and diverse Overpass queries. Third, we leverage a prompt-based approach to guide large language models in generating context-aware natural language questions, ensuring increasing inter-turn dependency across the dialogue. Finally, we implement a hybrid filtering strategy that combines manual annotation with model-assisted selection to validate alignment and correctness at scale. In addition to presenting the dataset, we evaluate the performance of several mainstream large language models and demonstrate that our end-to-end baseline model achieves competitive results. This work offers a new benchmark for studying executable semantic parsing and contextual understanding in map-based query tasks.
State-of-the-art large multi-modal models (LMMs) face challenges when processing high-resolution images, as these inputs are converted into enormous visual tokens, many of which are irrelevant to the downstream task. In this paper, we propose Multi-turn Grounding-based Policy Optimization (MGPO), an end-to-end reinforcement learning (RL) framework that enables LMMs to iteratively focus on key visual regions by automatically cropping sub-images, based on model-predicted grounding coordinates within a multi-turn conversation framework. Compared to supervised fine-tuning (SFT), which requires costly additional grounding annotations, our approach highlights that LMMs can emerge robust grounding abilities during the RL training process, leveraging only a binary reward function derived from the correctness of the final answer. Additionally, we observe that LMMs struggle to autonomously trigger visual grounding during the rollout process. To address this cold start problem, we design a multi-turn conversational template and restrict policy loss computation to model outputs generated across multiple dialogue rounds, thereby promoting stable optimization. Extensive experiments demonstrate that, when trained on standard visual-question-short answering data without grounding annotations, MGPO effectively elicits stronger grounding capabilities compared to GRPO, leading to 5.4% improvement on in-distribution MME-Realworld and 5.2% improvement on the challenging out-of-distribution (OOD) V* Bench. Notably, MGPO post-training on Qwen2.5-VL-7B with 21K samples surpasses OpenAI’s o1 and GPT-4o models on the OOD V* Bench.
I Don’t Need Solution. I Need Emotional Support : Empathetic LLMs based on Emotional Validation
PDF ↗Empathy plays a crucial role in prosocial behavior and supportive human interactions. According to emotional validation theory, effective empathetic conversations require observing and reflecting on the help-seeker’s situation before offering emotional support and guidance. While recent advancements in large language models (LLMs) have enabled fluent and coherent dialogue generation, our preliminary study reveals that existing LLMs struggle to generate emotional support response. Instead, they tend to offer repetitive solutions without sufficiently considering the emotional needs of help-seekers. To address this limitation, we propose EVA: empathetic LLMs with Emotional VAlidation. EVA enhances empathetic response generation through a two-stage training process: empathy acquisition and emotional validation alignment. For the emotional validation alignment, we introduce the Emotional Validation Aware Dataset (EVAD), which is annotated with levels of emotional validation theory as conversations progress. Additionally, we propose EVAEval, a novel evaluation metric designed to assess whether a model-generated response aligns with emotional validation theory. Experimental results demonstrate that the EVA method significantly improves empathetic response generation, achieving superior performance in both automatic and human evaluations. Furthermore, comprehensive analyses confirm that the EVA method effectively mitigates patterned responses while ensuring adherence to emotional validation principles.
Large Language Model agents can seeminglyplan and act, yet their language use is oftentreated primarily as an interface for instructingactions and reporting results. We argue that thisframing is one important cause of recurrent coordination failures in human-facing and multiagent settings, including ungrounded assumptions, silent goal misalignment, brittle protocoladherence, and failures to maintain or updateshared dialogue state over time, a limitation previously linked to the absence of explicit common ground tracking in collaborative systems(Geib et al., 2022). Drawing from classical dialogue system research on joint action, commonground, grounding, repair, and incremental processing, we re-frame dialogue as part of theplanning loop itself (rather than its output). Wedistill this re-framing into concrete implicationsfor agentic architecture and evaluation, including explicit representations of shared commitments, clarification as a first class action available to the policy, and process metrics that approximate grounding behavior, repair, and commitment formation rather than task completionalone. We lastly discuss how dialogue-centeredrequirements can inform standards and governance for safe deployment of agentic systems.
Full-Duplex-Bench-v2: A Multi-Turn Evaluation Framework for Duplex Dialogue Systems with an Automated Examiner
PDF ↗While full-duplex speech agents enable natural, low-latency interaction by speaking and listening simultaneously, their consistency and task performance in multi-turn settings remain underexplored. We introduce Full-Duplex-Bench-v2 (FDB-v2), a streaming framework that integrates with an automated examiner that enforces staged goals under two pacing setups (Fast vs. Slow). FDB-v2 covers four task families—Daily, Correction, Entity Tracking, and Safety—and reports turn-taking fluency, multi-turn instruction following, and task-specific competence. The framework is extensible, supporting both commercial APIs and open-source models. When we test full-duplex systems with FDB-v2, they often get confused when people talk at the same time, struggle to handle corrections smoothly, and sometimes lose track of who or what is being talked about. Through an open-source, standardized streaming protocol and a task set, FDB-v2 makes it easy to extend to new task families, allowing the community to tailor and accelerate evaluation of multi-turn full-duplex systems.
CIG: Measuring Conversational Information Gain in Deliberative Dialogues with Semantic Memory Dynamics
PDF ↗Measuring the quality of public deliberation requires evaluating not only civility or argument structure, but also the informational progress of a conversation. We introduce a framework for Conversational Information Gain (CIG) that evaluates each utterance in terms of how it advances collective understanding of the target topic. To operationalize CIG, we model an evolving semantic memory of the discussion: the system extracts atomic claims from utterances and incrementally consolidates them into a structured memory state. Using this memory, we score each utterance along three interpretable dimensions: Novelty, Relevance, and Implication Scope. We annotate 80 segments from two moderated deliberative settings (TV debates and community discussions) with these dimensions and show that memory-derived dynamics (e.g., the number of claim updates) correlate more strongly with human-perceived CIG than traditional heuristics such as utterance length or TF–IDF. We develop effective LLM-based CIG predictors paving the way for information-focused conversation quality analysis in dialogues and deliberative success.
Power differences shape human communication through well-documented socio-cognitive effects, including language coordination, pronoun usage, authority bias, and harmful compliance. We examine whether large language models (LLMs) exhibit similar behaviors when assigned high- or low-status personas. Using personas from diverse professions, we simulate multi-turn, power-asymmetric dialogues (e.g., principal–teacher, justice–lawyer) and measure (i) linguistic coordination, (ii) pronoun usage, (iii) persuasion success, and (iv) compliance with unsafe requests. Our results show that LLMs show key socio-cognitive effects of power, albeit with nuances and variability, linking simulated interactions to both desirable and unsafe behaviors.
With the growing popularity of virtual character platforms like Character.AI, users are increasingly turning to role-playing agents for emotional support in daily life. Yet existing research mainly focuses on character consistency in fictional or game-based scenarios, overlooking user-centered interactions such as companionship and psychological support. To bridge this gap, we propose Emotionally Supportive Role-Playing (ESRP), a framework designed to align role-playing with real-world user scenarios and emotional needs. We focus on typical users of these platforms, i.e., anime enthusiasts—including students, office workers, freelancers, and self-employed individuals—and design scenario-based questions that reflect their everyday struggles such as work stress and social loneliness. Through a two-round data collection involving 40 anime fans and 10 Large Language Models (LLMs), we build ChatAnime: the first ESRP dataset with 2,400 human-written and 24,000 LLM-generated responses, supported by over 132,000 fine-grained human annotations. We also provide the ESRP evaluation framework featuring 9 fine-grained metrics across three dimensions: basic dialogue, role-playing and emotional support, along with an overall metric for diversity. Experimental results under our evaluation setting show that top-performing LLMs surpass anime fans in role-playing and emotional support, while humans still lead in diversity.
Limited access to mental healthcare, extended wait times, and increasing capabilities of Large Language Models (LLMs) has led individuals to turn to LLMs for fulfilling their mental health needs. However, examining the multi-turn mental health conversation capabilities of LLMs remains under-explored. Existing evaluation frameworks typically focus on diagnostic accuracy and win-rates and often overlook alignment with patient-specific goals, values, and personalities required for meaningful conversations. To address this, we introduce MedAgent, a novel framework for synthetically generating realistic, multi-turn mental health sensemaking conversations and use it to create the Mental Health Sensemaking Dialogue (MHSD) dataset, comprising over 2,200 patient–LLM conversations. Additionally, we present MultiSenseEval, a holistic framework to evaluate the multi-turn conversation abilities of LLMs in healthcare settings using human-centric criteria. Our findings reveal that frontier reasoning models yield below-par performance for patient-centric communication and struggle at precise ("hard") diagnostic capabilities with average accuracy of ~31%. Additionally, we observed variation in model performance based on patient’s persona and performance drop with increasing turns in the conversation. Our work provides a comprehensive synthetic data generation framework, a dataset and evaluation framework for assessing LLMs in multi-turn mental health conversations.
Emotional Support Conversation requires not only affective expression but also grounded instrumental support to provide trustworthy guidance. However, existing ESC systems and benchmarks largely focus on affective support in text-only settings, overlooking how external tools can enable factual grounding and reduce hallucination in multi-turn emotional support. We introduce **TEA-Bench**, the first interactive benchmark for evaluating tool-augmented agents in ESC, featuring realistic emotional scenarios, an MCP-style tool environment, and process-level metrics that jointly assess the quality and factual grounding of emotional support. Experiments on nine LLMs show that tool augmentation generally improves emotional support quality and reduces hallucination, but the gains are strongly capacity-dependent: stronger models use tools more selectively and effectively, while weaker models benefit only marginally. We further release **TEA-Dialog**, a dataset of tool-enhanced ESC dialogues, and find that supervised fine-tuning improves in-distribution support but generalizes poorly. Our results underscore the importance of tool use in building reliable emotional support agents.
Evaluating speech generation still relies heavily on human judgments, such as Mean Opinion Score (MOS), which are expensive, subjective, and difficult to reproduce at scale. While a few recent studies have begun to explore AudioLLM-based judge models, existing efforts typically target only a narrow set of scenarios (e.g., utterance-level quality or single-turn dialogue) and provide limited coverage of diverse speech generation tasks and evaluation dimensions. In this work, we propose UniSRM, a unified speech reward model that can support multi-dimensional, interpretable reward signals with reliable reasoning. To support training and evaluation, we introduce UniSRM-Data and UniSRM-Bench, covering speech evaluation tasks from utterance-level quality to context-level coherence. Based on this dataset, we present the unified speech reward model, UniSRM, with a two-stage pipeline that enables reasoning-based fine-grained assessment. Furthermore, we introduce Reasoning-Consistent Rewards to improve the reliability of the reasoning process. Experiments show that UniSRM delivers more reliable and human-aligned judgments across a broad range of speech evaluation tasks, offering a practical foundation for scalable and unified evaluation of speech quality.
Language models are increasingly deployed in interactive settings where users reason about facts over time rather than in isolation.In such scenarios, correct behavior requires models to maintain and update implicit temporal assumptions established earlier in a conversation. We study this challenge through the lens of temporal scope stability: the ability to preserve, override, or transfer time-scoped factual context across dialogue turns. We introduce ChronoScope, a large-scale diagnostic benchmark designed to isolate temporal scope behavior in controlled multi-turn interactions, comprising over one million deterministically generated question chains grounded in Wikidata. ChronoScope evaluates whether models can correctly retain inferred temporal scope when follow-up questions omit explicit time references, spanning implicit carryover, explicit scope switching, cross-entity transfer, and longer temporal trajectories.Through extensive evaluation of state-of-the-art language models, we find that temporal scope stability is frequently violated in controlled multi-turn settings, with models often drifting toward present-day assumptions despite correct underlying knowledge.These failures intensify with interaction length and persist even under oracle context conditions, revealing a gap between single-turn factual accuracy and coherent temporal reasoning under sequential interaction.We make our dataset and evaluation suite publicly available at https://github.com/yashkumaratri/ChronoScope.