论文检索

Ryo Kishino, Yusuke Takase, Momose Oyama, Hiroaki Yamagiwa, Hidetoshi Shimodaira

Log-likelihood vectors define a common space for comparing language models as probability distributions, enabling unified comparisons across heterogeneous settings. We extend this framework to training checkpoints and intermediate layers, and establish a consistent scale for KL divergence across pretraining, model size, random seeds, quantization, fine-tuning, and layers. Analysis of Pythia pretraining trajectories further shows that changes in log-likelihood space, as measured by the scaling behavior of KL divergence, are much smaller than in weight space, resulting in subdiffusive learning trajectories and early stabilization of language-model behavior despite weight drift.

Chenxi Zhou, Pengfei Cao, Jiang Li, Bohan Yu, Jinyu Ye, Jun Zhao, Kang Liu

Post-Training Quantization (PTQ) is critical for the efficient deployment of Large Language Models (LLMs). While 4-bit quantization is widely regarded as an optimal trade-off, reducing the precision to 2-bit usually triggers a catastrophic “performance cliff.” It remains unclear whether the underlying mechanisms differ fundamentally. Consequently, we conduct a systematic mechanistic analysis, revealing two qualitatively distinct failure modes: Signal Degradation, where the computational patterns remain intact but information precision is impaired by cumulative error; and Computation Collapse, where key components fail to function, preventing correct information processing and destroying the signal in the early layers. Guided by this diagnosis, we conduct mechanism-aware interventions, demonstrating that targeted, training-free repair can mitigate Signal Degradation, but remains ineffective for Computation Collapse. Our findings provide a systematic diagnostic framework for PTQ failures and suggest that addressing Computation Collapse requires structural reconstruction rather than mere compensation.

Xiaoyong Mei, Tingting Zuo, Da Chen, Guangyu Hu, Xiangyu Wen, Chao Duan, Mingyan Zhang, Fudan Zheng

Multi-role dialogue summarization requires modeling complex interactions among multiple speakers while preserving role-specific information and factual consistency. However, most existing methods optimize for automatic metrics such as ROUGE and BERTScore, which favor surface-level imitation of references rather than genuine gains in faithfulness or alignment with human preferences. We propose a novel framework that couples explicit cognitive-style reasoning with reward-based optimization for multi-role dialogue summarization. Our method first distills structured reasoning traces (e.g., step-by-step inferences and intermediate reflections) from a large teacher model and uses them as auxiliary supervision to initialize a reasoning-aware summarizer via staged supervised fine-tuning. It then applies GRPO with a dual-principle reward that blends metric-based signals with human-aligned criteria targeting key information coverage, implicit inference, factual faithfulness, and conciseness. Experiments on multilingual multi-role dialogue benchmarks show that our method matches strong baselines on ROUGE and BERTScore. Specifically, results on CSDS confirm the framework’s stability in semantic consistency, while in-depth analysis on SAMSum demonstrates clear gains in factual faithfulness and model-based preference alignment. These findings underscore the value of reasoning-aware and preference-aware training for reliable dialogue summarization. Code will be made accessible upon acceptance, checkpoints and datasets are now available at https://huggingface.co/NebulaPixel.

Yifei Gong, Xing Wu, Wenda Liu, Tukang

Computer-Aided Design (CAD) is an expert-level task that relies on long-horizon reasoning and coherent modeling actions. Large Language Models (LLMs) have shown remarkable advancements in enabling language agents to tackle real-world tasks. Notably, there has been no investigation into how tool-using LLMs optimally interact with CAD engines, hindering the emergence of LLM-based agentic text-to-CAD modeling systems. We propose ToolCAD, a novel agentic CAD framework deploying LLMs as tool-using agents for text-to-CAD generation. Furthermore, we introduce an interactive CAD modeling gym to rollout reasoning and tool-augmented interaction trajectories with the CAD engine, incorporating hybrid feedback and human supervision. Meanwhile, an end-to-end post-training strategy is presented to enable the LLM agent to elicit refined CAD Modeling Chain of Thought (CAD-CoT) and evolve into proficient CAD tool-using agents via online curriculum reinforcement learning. Our findings demonstrate ToolCAD fills the gap in adopting and training open-source LLMs for CAD tool-using agents, enabling them to perform comparably to proprietary models, paving the way for more accessible and robust autonomous text-to-CAD modeling systems.

Zhuoqun Li, Zhaopei Huang, Wenxuan Wang, Qin Jin

Understanding multimodal user personas in long-term dialogues is essential for building personalized and human-like dialogue systems. However, existing datasets suffer from limited persona diversity and static, overly simplified settings, making them insufficient for capturing the complexity of real-world interactions. To address these limitations, we introduce LongMP-Bench, a benchmark designed to evaluate the capabilities of models in understanding evolving user personas within long-term multimodal dialogues. We present a multi-step, scalable data construction pipeline that generates long-term interaction records centered around multimodal personas, followed by human refinement for quality assurance. The resulting dataset contains long conversations from 150 users, each exhibiting visual consistency and dynamic persona development over time. Built on this dataset, we define a suite of tasks to comprehensively assess models’ ability to track persona evolution, integrate visual and textual inputs, and apply persona understanding in realistic dialogue scenarios. Extensive experiments on LongMP-Bench highlight the substantial challenges in multimodal persona understanding, especially in tracking persona shifts and leveraging multimodal context effectively. We will release our benchmark and code to facilitate future research in multimodal and personalized dialogue systems.

Zhe Chen, Jiaao Yu, Honglin Li

Length-controllable text generation (LCTG) is essential for tasks like text summarization and report generation. However, large language models (LLMs) have limited awareness of output length, so precise control over the length of generated text remains a significant challenge. Most existing methods focus on prompt-based frameworks, position encoding, and reinforcement learning for model training. These approaches may affect semantic quality, and struggle to maintain consistent length control across different models and tasks. In this paper, we propose DcLM, a model-agnostic approach that introduces dynamic length markers to guide length-controllable outputs. During training, the model leverages these markers as in-context information, without learning to generate them. At inference time, an external word counter and injected length information guide the model to produce outputs of accurate lengths. We evaluate our method across multiple datasets, and the experimental results demonstrate that DcLM significantly reduces length deviation, showcasing its robust generalization ability across various length scales and tasks.

Yanru Huo, Ziyue Jiang, Zuoli Tang, Qingyang Hong, Zhou Zhao

While Diffusion Transformers (DiT) have advanced non-autoregressive (NAR) speech synthesis, their high computational demands remain an obvious limitation. Existing DiT-based text-to-speech (TTS) model acceleration approaches predominantly focus on reducing sampling steps through distillation techniques, yet they remain constrained by training costs. We introduce DiTReducio, a training-free acceleration framework that compresses computations in DiT-based TTS models through a progressive calibration process. We propose two compression methods, Temporal Skipping and Branch Skipping, to eliminate redundant computations during inference. Moreover, based on two characteristic attention patterns identified within DiT layers, we devise a pattern-guided strategy to selectively apply the compression methods. Our method allows flexible modulation between generation quality and computational efficiency through adjustable compression thresholds. Experimental evaluations conducted on F5-TTS and MegaTTS 3 demonstrate that DiTReducio achieves a 75.4% reduction in FLOPs and improves the Real-Time Factor (RTF) by 37.1%, while preserving generation quality. The code is available at https://github.com/MM-Speech/DiTReducio.

Jiazheng Zhang, Ziche Fu, Zhiheng Xi, Wenqing Jing, Mingxu Chai, Wei He, Guoqiang Zhang, Chenghao Fan, Chenxin An, Wenxiang Chen 等

Verifiers have been demonstrated to enhance LLM reasoning via test-time scaling (TTS). Yet, they face significant challenges in complex domains. Error propagation from incorrect intermediate reasoning can lead to false positives for seemingly plausible solutions, while lacking external grounding makes verifiers unreliable on computation or knowledge-intensive tasks. To address these challenges, we propose Agentic Verifier, a framework that transforms reward modeling into a multi-turn, tool-augmented deliberative process. We introduce complementary forward and backward agents: one traces solutions from premises to conclusions, while the other re-checks conclusions against their underlying premises. This bidirectional process enables a comprehensive, reliable, and interpretable assessment of solutions. To facilitate practical deployment, we propose AgentV-RL. Through proactive exploration and reinforcement learning, the verifier autonomously interleaves tool-use with internal reasoning. Extensive experiments show that Agentic Verifier yields consistent performance gains under both parallel and sequential TTS. Notably, our 4B variant surpasses state-of-the-art ORMs by 25.2%, positioning it as a promising paradigm for agentic reward modeling.

Yiwen Qiu, Linjuan Wu, Yizhou Liu, Yuchen Yan, Jin Ma, Xu Tan, Yao Hu, Daoxin Zhang, Wenqi Zhang, Weiming Lu 等

Large language models have achieved remarkable progress on complex reasoning tasks. However, they often implicitly fabricate information when inputs are incomplete, producing confident but unreliable conclusions—a failure mode we term ungrounded reasoning. We argue that this issue arises not from insufficient reasoning capability, but from the lack of inferential boundary awareness—the ability to recognize when the necessary premises for valid inference are missing. To address this issue, we propose Grounded Reasoning via Interactive Reinforcement Learning (GRIL), a multi-turn reinforcement learning framework for grounded reasoning under incomplete information. GRIL decomposes the reasoning process into two stages: clarify and pause, which identifies whether the available information is sufficient, and grounded reasoning, which performs task solving once the necessary premises are established. We design stage-specific rewards to penalize hallucinations, enabling models to detect gaps, stop proactively, and resume reasoning after clarification. Experiments on GSM8K-Insufficient and MetaMATH-Insufficient show that GRIL significantly improves premise detection (up to 45%), leading to a 30% increase in task success while reducing average response length by over 20%. Additional analyses confirm robustness to noisy user responses and generalization to out-of-distribution tasks.

Yu Cui, Hang Fu, Sicheng Pan, Zhuoyu Sun, Yifei Liu, Yuhong Nie, Bo Ran, Baohan Huang, Xufeng Zhang, Haibin Zhang 等

Existing research on generative AI security is primarily driven by mutually reinforcing attack and defense methodologies grounded in empirical experience. This dynamic frequently gives rise to previously unknown attacks that can circumvent current detection and prevention. This necessitates the continual updating of security mechanisms. Constructing generative AI with provable security and theoretically controllable risk is therefore necessary. Consensus Sampling (CS) is a promising algorithm toward provably secure AI. It controls risk by leveraging overlap in model output probabilities. However, we find that CS relies on frequent abstention to avoid unsafe outputs, which reduces utility. Moreover, CS becomes highly vulnerable when unsafe models are maliciously manipulated. To address these issues, we propose a new primitive called Reliable Consensus Sampling (RCS), that traces acceptance probability to tolerate extreme adversarial behaviors, improving robustness. RCS also eliminates the need for abstention entirely. We further develop a feedback algorithm to continuously and dynamically enhance the safety of RCS. We provide theoretical guarantees that RCS maintains a controllable risk threshold. Extensive experiments show that RCS significantly improves robustness and utility while maintaining latency comparable to CS. We hope this work contributes to the development of provably secure generative AI. Our code is available at https://github.com/cuiyu-ai/RCS.

Tian Xueyun, Wei Li, Bingbing Xu, Heng Dong, Yuanzhuo Wang, Huawei Shen

Recent Omni-multimodal Large Language Models show promise in unified audio, vision, and text modeling. However, streaming audio-video understanding remains challenging, as existing approaches suffer from disjointed capabilities: they typically exhibit incomplete modality support or lack autonomous proactive monitoring. To address this, we present ROMA, **a real-time omni-multimodal assistant for unified reactive and proactive interaction**. ROMA processes continuous inputs as synchronized multimodal units, aligning dense audio with discrete video frames to handle granularity mismatches. For online decision-making, we introduce a lightweight *speak head* that decouples response initiation from generation to ensure precise triggering without task conflict. We train ROMA with a curated streaming dataset and a two-stage curriculum that progressively optimizes for streaming format adaptation and proactive responsiveness. To standardize the fragmented evaluation landscape, we reorganize diverse benchmarks into a unified suite covering both proactive (alert, narration) and reactive (QA) settings. Extensive experiments across 12 benchmarks demonstrate ROMA achieves state-of-the-art performance on proactive tasks while competitive in reactive settings, validating its robustness in unified real-time omni-multimodal understanding. Code and benchmark are available [here](https://eureka-maggie.github.io/ROMA_show/).

Amaia Cardiel, Eloi Zablocki, Elias Ramzi, Eric Gaussier

Large language models (LLMs) are increasingly used as reasoning engines in autonomous driving, yet their decision-making remains opaque. We propose to study their decision process through counterfactual explanations, which identify the minimal semantic changes to a scene description required to alter a driving plan.We introduce DRIV-EX, a method that leverages gradient-based optimization on continuous embeddings to identify the input shifts required to flip the model’s decision. Crucially, to avoid the incoherent text typical of unconstrained continuous optimization, DRIV-EX uses these optimized embeddings solely as a semantic guide: they are used to bias a controlled decoding process that re-generates the original scene description. This approach effectively steers the generation toward the counterfactual target while guaranteeing the linguistic fluency, domain validity, and proximity to the original input essential for interpretability.Evaluated using the LC-LLM planner on the textual highD dataset, DRIV-EX generates valid, fluent counterfactuals more reliably than existing baselines. It successfully exposes latent biases and provides concrete insights to improve the robustness of LLM-based driving agents. The code is available at "https://github.com/Amaia-CARDIEL/DRIV_EX".

Adam Ishay, Joohyung Lee

Recent large language models (LLMs) have achieved impressive reasoning milestones but continue to struggle with high computational costs, logical inconsistencies, and sharp performance degradation on high-complexity problems. While neuro-symbolic methods attempt to mitigate these issues by coupling LLMs with symbolic reasoners, existing approaches typically rely on monotonic logics (e.g., SMT) that cannot represent defeasible reasoning—essential components of human cognition. We present "LLM+ASP," a framework that translates natural language into Answer Set Programming (ASP), a nonmonotonic formalism based on stable model semantics. Unlike prior LLM+ASP approaches that require manually authored knowledge modules, domain-specific prompts, or evaluation restricted to single problem classes, our framework operates without any per-task engineering and applies uniformly across diverse reasoning tasks. Our system utilizes an automated self-correction loop where structured feedback from the ASP solver enables iterative refinement. Evaluating across six diverse benchmarks, we demonstrate that: (1) stable model semantics allow LLMs to naturally express default rules and exceptions, outperforming SMT-based alternatives by significant margins on nonmonotonic tasks; (2) iterative self-correction is the primary driver of performance, effectively replacing the need for handcrafted domain knowledge; (3) compact in-context reference guides substantially outperform verbose documentation, revealing a “context rot" phenomenon where excessive context hinders constraint adherence.

Hyunji Min, Sangwon Jung, Junyoung Sung, Dosung Lee, Leekyeung Han, Paul Hongsuck Seo

Large language models (LLMs) have evolved from pure text generators into interactive agents capable of invoking external tools. However, LLM agents still struggle with goal-oriented queries, which require decomposing high-level objectives into sequences of interdependent API calls with accurate planning and execution. Current approaches rely on zero-shot evaluation due to the absence of training data; while proprietary models such as GPT-4 exhibit strong reasoning capabilities, smaller open-source models remain ineffective at complex tool use. To address this limitation, we propose a novel training framework GOAT, that enables fine-tuning LLM agents without human annotation. GOAT automatically synthesizes goal-oriented API execution data from API documents using a novel call-first generation paradigm, that constructs training data based on executed API call sequences. Through extensive experiments, we show that GOAT-trained agents achieve state-of-the-art performance across multiple existing goal-oriented benchmarks. In addition, we introduce GOATBench, a new goal-oriented API execution benchmark, and demonstrate that agents trained with GOAT also excel in this setting. These results highlight GOAT as a practical path toward building robust open-source LLM agents capable of complex reasoning and tool use.

Guoming Ling, Zhongzhan Huang, Yupei Lin, Junxin Li, Shanshan Zhong, Hefeng Wu, Liang Lin

Chain-of-Thought reasoning has significantly enhanced the problem-solving capabilities of Large Language Models. Unfortunately, current models generate reasoning steps sequentially without foresight, often becoming trapped in suboptimal reasoning paths with redundant steps. In contrast, we introduce Neural Chain-of-Thought Search (NCoTS), a framework that reformulates reasoning as a dynamic search for the optimal thinking strategy. By quantitatively characterizing the solution space, we reveal the existence of sparse superior reasoning paths that are simultaneously more accurate and concise than standard outputs. Our method actively navigates towards these paths by evaluating candidate reasoning operators using a dual-factor heuristic that optimizes for both correctness and computational cost. Consequently, NCoTS achieves a Pareto improvement across diverse reasoning benchmarks, boosting accuracy by over 3.5% while reducing generation length by over 22%. We will make our code and data publicly available.

Junzhe Zhou, Fulin Lin, Tairan Cheng, Shaowen Chen, Hongwei Wang

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) yet suffers from a mismatch between coarse retrieval granularity and fine-grained generation needs. Specifically, coarse-grained passages inherently conflate valid context with intra-passage noise that semantic retrieval often fails to filter. Existing alignment strategies, typically relying on discrete reranking, struggle to address this granularity mismatch or effectively balance external evidence with internal knowledge. To bridge this gap, we propose **AED-RAG**, a framework that synergizes discrete retrieval with continuous **A**daptive **E**nsemble **D**ecoding. Specifically, we fine-tune a utility predictor using contrastive perplexity to discern the information density differences between unstructured narrative passages and structured knowledge triplets. During inference, this predictor projects passages, triplets, and the model’s parametric memory into a unified probability space, enabling a soft, token-level fusion that dynamically optimizes information gain. Extensive experiments on four open-domain QA benchmarks demonstrate that AED-RAG significantly outperforms competitive baselines, underscoring the effectiveness of integrating multi-granular contexts.

Xiang Li, Runhai Jiao, Ruifan Li, Dongnan Wu, Ruojiao Qiao, Lei Liu

Reinforcement learning, with its interpretable path reasoning, has emerged as a promising paradigm for multi-hop question answering over knowledge graphs. However, existing approaches suffer from two inherent limitations: (1) lacking effective intermediate guidance, agents often fall into aimless exploration when confronted with complex multi-hop questions; and (2) policy networks focus on local neighborhood information, making it difficult to anticipate the long-term consequences of decisions. To address these challenges, we propose a Progressive Planning and Reinforced Reasoning (PPRR) framework. Specifically, we introduce large language models as multi-hop reasoning planners, converting decomposed sub-question sequences into stepwise decision guidance and thereby granting the agent human-like, step-by-step problem-solving capabilities. In addition, we design a structure-aware lookahead policy network, which explicitly models inter-node dependencies along the multi-hop reasoning process and performs lookahead value evaluations for candidate actions, thereby enhancing the agent’s global state awareness and decision foresight in complex environments. Finally, we conducted extensive experiments on four public multi-hop question answering benchmarks and one domain-specific dataset. The results demonstrate that our framework surpasses state-of-the-art methods while demonstrating strong generalization.

Chaewon Heo, Cheyon Jin, Yohan Jo

As emotional support chatbots have recently gained significant traction across both research and industry, a common evaluation strategy has emerged: use help-seeker simulators to interact with supporter chatbots. However, current simulators suffer from two critical limitations: (1) they fail to capture the behavioral diversity of real-world seekers, often portraying them as overly cooperative, and (2) they lack the controllability required to simulate specific seeker profiles. To address these challenges, we present a controllable seeker simulator driven by nine psychological and linguistic features that underpin seeker behavior. Using authentic Reddit conversations, we train our model via a Mixture-of-Experts (MoE) architecture, which effectively differentiates diverse seeker behaviors into specialized parameter subspaces, thereby enhancing fine-grained controllability. Our simulator achieves superior profile adherence and behavioral diversity compared to existing approaches. Furthermore, evaluating 7 prominent supporter models with our system uncovers previously obscured performance degradations. These findings underscore the utility of our framework in providing a more faithful and stress-tested evaluation for emotional support chatbots.

Finn Schmidt, Jan Philip Wahle, Terry Ruas, Bela Gipp

Automatic evaluation metrics are central to the development of machine translation systems, yet their robustness under domain shift remains unclear. Most metrics are developed on the Workshop on Machine Translation (WMT) benchmarks, raising concerns about their robustness to unseen domains. Prior studies that analyze unseen domains vary translation systems, annotators, or evaluation conditions, confounding domain effects with human annotation noise.To address these biases, we introduce a systematic multi-annotator **C**ross-**D**omain **E**rror-**S**pan-**A**nnotation dataset (CD-ESA), comprising 18.8k human error span annotations across three language pairs, where we fix annotators within each language pair and evaluate translations of the same six translation systems across one seen news domain and two unseen technical domains. Using this dataset, we first find that automatic metrics appear surprisingly robust to domain-shifts at the segment level (up to 0.69 agreement), but this robustness largely disappears once we account for human label variation. Averaging annotations increases inter-annotator agreement by up to +0.11. Metrics struggle on the unseen chemical domain compared to humans (inter-annotator agreement of 0.78–0.83 vs. 0.96).We recommend comparing metric–human agreement against inter-annotator agreement, rather than comparing raw metric–human agreement alone, when evaluating across different domains.

Hugh Mee Wong, Rick Nouwen, Albert Gatt

Multiple-choice question answering (MCQA) is easy to evaluate but adds a meta-task: models must both solve the problem and output the symbol that represents the answer, conflating reasoning errors with symbol-binding failures. We study how language models implement MCQA internally using representational analyses (PCA, linear probes) as well as causal interventions. We find that option-boundary (newline) residual states often contain strong linearly decodable signals related to per-option correctness. Winner-identity probing reveals a two-stage progression: the winning content position becomes decodable immediately after the final option is processed, while the output symbol is represented closer to the answer emission position. Tests under symbol and content permutations support a two-stage mechanism in which models first select a winner in content space and then bind or route that winner to the appropriate symbol to emit.