The Model Context Protocol (MCP) is rapidly emerging as a pivotal open standard, designed to enhance agent-tool integration and interoperability, and is positioned to unlock a new era of powerful, interconnected, and genuinely utilitarian agentic AI. However, despite MCP's growing adoption, existing benchmarks often fail to capture real-world agent performance within this new paradigm, leading to a distorted perception of their true operational value and an inability to reliably differentiate proficiencies. To bridge this critical evaluation gap, we introduce MCP-AgentBench—a comprehensive benchmark specifically engineered to rigorously assess language agent capabilities in MCP-mediated tool interactions. Core contributions of MCP-AgentBench include: the establishment of a robust MCP testbed comprising 33 operational servers with 188 distinct tools; the development of a benchmark featuring 600 systematically designed queries distributed across 6 distinct categories of varying interaction complexity; and the introduction of MCP-Eval, a novel outcome-oriented evaluation methodology prioritizing real-world task success. Through extensive empirical evaluation of leading language agents, we provide foundational insights. MCP-AgentBench aims to equip the research community with a standardized and reliable framework to build, validate, and advance agents capable of fully leveraging MCP's transformative benefits, thereby accelerating progress toward truly capable and interoperable AI systems.
论文检索
输入标题、作者或关键词，从 100,903 篇学术成果中精准定位
Extracting Events Like Code: A Multi-Agent Programming Framework for Zero-Shot Event Extraction
PDF ↗Zero-shot event extraction (ZSEE) remains a significant challenge for large language models (LLMs) due to the need for complex reasoning and domain-specific understanding. Direct prompting often yields incomplete or structurally invalid outputs—such as misclassified triggers, missing arguments, and schema violations. To address these limitations, we present Agent-Event-Coder (AEC), a novel multi-agent framework that treats event extraction like software engineering: as a structured, iterative code-generation process. AEC decomposes ZSEE into specialized subtasks—retrieval, planning, coding, and verification—each handled by a dedicated LLM agent. Event schemas are represented as executable class definitions, enabling deterministic validation and precise feedback via a verification agent. This programming-inspired approach allows for systematic disambiguation and schema enforcement through iterative refinement. By leveraging collaborative agent workflows, AEC enables LLMs to produce precise, complete, and schema-consistent extractions in zero-shot settings. Experiments across five diverse domains and six LLMs demonstrate that AEC consistently outperforms prior zero-shot baselines, showcasing the power of treating event extraction like code generation.
Seeing Is Believing: Rich-Context Hallucination Detection for MLLMs via Backward Visual Grounding
PDF ↗Multimodal Large Language Models (MLLMs) have unlocked powerful cross-modal capabilities, but still significantly suffer from hallucinations. As such, accurate detection of hallucinations in MLLMs is imperative for ensuring their reliability in practical applications. To this end, guided by the principle of “Seeing is Believing”, we introduce VBackChecker, a novel reference-free hallucination detection framework that verifies the consistency of MLLM-generated responses with visual inputs, by leveraging a pixel-level Grounding LLM equipped with reasoning and referring segmentation capabilities. This referencefree framework not only effectively handles rich-context scenarios, but also offers interpretability. To facilitate this, an innovative pipeline is accordingly designed for generating instruction-tuning data (R-Instruct), featuring richcontext descriptions, grounding masks, and hard negative samples. We further establish R 2 -HalBench, a new hallucination benchmark for MLLMs, which, unlike previous benchmarks, encompasses real-world, rich-context descriptions from 18 MLLMs with high-quality annotations, spanning diverse object-, attribute-, and relationship-level details. VBackChecker outperforms prior complex frameworks and achieves state-of-the-art performance on R^2 -HalBench, even rivaling GPT-4o’s capabilities in hallucination detection. It also surpasses prior methods in the pixel-level grounding task, achieving over a 10% improvement.
Knowledge editing (KE) has emerged as an effective approach for updating factual information in large language models (LLMs) without the need for full retraining. Most of the existing methods for addressing the "ripple effect" in KE adopt a chain-structured reasoning process, making them vulnerable to error accumulation from early incorrect steps. Moreover, their conflict detection mechanisms are often susceptible to the LLM's inherent confirmation bias, further undermining the reliability of the editing process. To overcome these challenges, we propose Tree of Editing (ToE), a tree-structured, retrieval-enhanced knowledge editing framework designed to support robust reasoning under factual updates. ToE expands reasoning paths using a breadth-first strategy combined with score-guided beam search, enabling diverse and error-tolerant inference. Besides, we introduce an observer to objectively update knowledge, avoiding the bias caused by LLMs' over-confidence. Experimental results on two benchmarks, namely MQuAKE-CF (targeting ripple-aware editing) and DUNE (free-form editing), demonstrate that ToE framework significantly outperforms existing methods.
Beyond Superficial Forgetting: Thorough Unlearning Through Knowledge Density Estimation and Block Re-Insertion
PDF ↗Machine unlearning, which selectively removes harmful knowledge from a pre-trained model without retraining from scratch, is crucial for addressing privacy, regulatory compliance, and ethical concerns in Large Language Models (LLMs). However, existing unlearning methods often struggle to thoroughly remove harmful knowledge, leaving residual harmful knowledge that can be easily recovered. To address these limitations, we propose Knowledge Density-Guided Unlearning via Blocks Reinsertion (KUnBR), a novel approach that first identifies layers with rich harmful knowledge and then thoroughly eliminates the harmful knowledge via re-insertion strategy. Our method introduces knowledge density estimation to quantify and locate layers containing the most harmful knowledge, enabling precise unlearning. Additionally, we design a layer re-insertion strategy that extracts and re-inserts harmful knowledge-rich layers into the original LLM, bypassing gradient obstruction caused by cover layers and ensuring effective gradient propagation during unlearning. Extensive experiments conducted on several unlearning and general capability benchmarks demonstrate that KUnBR achieves state-of-the-art forgetting performance while maintaining model utility.
Beware of Reasoning Overconfidence: Pitfalls in the Reasoning Process for Multi-solution Tasks
PDF ↗Large Language Models (LLMs) excel in reasoning tasks requiring a single correct answer, but they perform poorly in multi-solution tasks that require generating comprehensive and diverse answers. We attribute this limitation to reasoning overconfidence: a tendency to express undue certainty in an incomplete solution set. To examine the effect, we introduce MuSoBench, a benchmark of multi-solution problems. Experiments show that the conventional short chain-of-thought (Short-CoT) prompting paradigm exhibits pronounced overconfidence, whereas the emerging long chain-of-thought (Long-CoT) approach mitigates it through iterative exploration and self-reflection. We further characterise observable behaviours and influential factors. To probe the underlying cause, we propose the cognitive-rigidity hypothesis, which posits that overconfidence arises when the reasoning process prematurely converges on a narrow set of thought paths. An attention-entropy analysis offers preliminary support for this view. These findings provide tools for assessing the completeness of LLM reasoning and highlight the need to move evaluation beyond single-answer accuracy toward comprehensive exploration.
Recent advances in large language models (LLMs) have broadened their applicability across diverse tasks, yet specialized domains still require targeted post-training. Among existing methods, Group Relative Policy Optimization (GRPO) stands out for its efficiency, leveraging groupwise relative rewards while avoiding costly value function learning. However, GRPO treats candidate responses as independent, overlooking semantic interactions such as complementarity and contradiction. To address this challenge, we first introduce a Structural Causal Model (SCM) that reveals hidden dependencies among candidate responses induced by conditioning on a final integrated output, forming a collider structure. Then, our causal analysis leads to two insights: (1) projecting responses onto a causally-informed subspace improves prediction quality, and (2) this projection yields a better baseline than query-only conditioning. Building on these insights, we propose Group Causal Policy Optimization (GCPO), which integrates causal structure into optimization through two key components: a causally-informed reward adjustment and a novel KL-regularization term that aligns the policy with a causally-projected reference distribution. Comprehensive experimental evaluations on various benchmarks demonstrate that GCPO consistently surpasses existing methods.
Multi-Agent Debate (MAD) is an emerging paradigm that leverages the reasoning abilities of Large Language Models (LLMs) by encouraging them to collaboratively solve problems through human-like discussions. However, current MAD methods typically constrain agents to follow fixed discussion pipelines, repeatedly applying the same discussion act for a predetermined number of rounds, which limits their effectiveness and adaptability in complex and diverse tasks. To address this limitation, we propose Analyze–Compose–Execute (ACE), a novel debate framework in which agents dynamically execute the discussion actions according to the dialogue context. By analyzing the current responses of agents, ACE selects appropriate acts from a predefined Atomic Discussion Acts Library (ADAL), which are composed into a discussion action to be executed in the next round, to enable truly dynamic debate. We conduct extensive experiments on the challenging benchmark Big-Bench Hard (BBH) benchmark. ACE achieves state-of-the- art results on 17 out of 23 tasks, with an average performance gain of 8.5% across all tasks, demonstrating the effectiveness and robustness of our approach.
Large language models (LLMs) are increasingly deployed with hierarchical instruction schemes, where certain instructions (e.g., system-level directives) are expected to take precedence over others (e.g., user messages). Yet, we lack a systematic understanding of how effectively these hierarchical control mechanisms work. We introduce a systematic evaluation framework based on constraint prioritization to assess how well LLMs enforce instruction hierarchies. Our experiments across six state-of-the-art LLMs reveal that models struggle with consistent instruction prioritization, even for simple formatting conflicts. We find that the widely-adopted system/user prompt separation fails to establish a reliable instruction hierarchy, and models exhibit strong inherent biases toward certain constraint types regardless of their priority designation. Interestingly, we also find that societal hierarchy framings (e.g., authority, expertise, consensus) show stronger influence on model behavior than system/user roles, suggesting that pretraining-derived social structures function as latent behavioral priors with potentially greater impact than post-training guardrails.
Speech-to-Speech (S2S) Large Language Models (LLMs) are foundational to natural human-computer interaction, enabling end-to-end spoken dialogue systems. However, evaluating these models remains a fundamental challenge. We propose SageLM, an end-to-end, multi-aspect, and explainable speech LLM for comprehensive S2S LLMs evaluation. First, unlike cascaded approaches that disregard acoustic features, SageLM jointly assesses both semantic and acoustic dimensions. Second, it leverages rationale-based supervision to enhance explainability and guide model learning, achieving superior alignment with evaluation outcomes compared to rule-based reinforcement learning methods. Third, we introduce SpeechFeedback, a synthetic preference dataset, and employ a two-stage training paradigm to mitigate the scarcity of speech preference data. Trained on both semantic and acoustic dimensions, SageLM achieves an 82.79% agreement rate with human evaluators, outperforming cascaded and SLM-based baselines by at least 7.42% and 26.20%, respectively.
Expert-Guided Prompting and Retrieval-Augmented Generation for Emergency Medical Service Question Answering
PDF ↗Large language models (LLMs) have shown promise in medical question answering, yet they often overlook the domain-specific expertise that professionals depend on-such as the clinical subject areas (e.g., trauma, airway) and the certification level (e.g., EMT, Paramedic). Existing approaches typically apply general-purpose prompting or retrieval strategies without leveraging this structured context, limiting performance in high-stakes settings. We address this gap with EMSQA, an 24.3K-question multiple-choice dataset spanning 10 clinical subject areas and 4 certification levels, accompanied by curated, subject area-aligned knowledge bases (40K documents and 2M tokens). Building on EMSQA, we introduce (i) Expert-CoT, a prompting strategy that conditions chain-of-thought (CoT) reasoning on specific clinical subject area and certification level, and (ii) ExpertRAG, a retrieval-augmented generation pipeline that grounds responses in subject area-aligned documents and real-world patient data. Experiments on 4 LLMs show that Expert-CoT improves up to 2.05% over vanilla CoT prompting. Additionally, combining Expert-CoT with ExpertRAG yields up to a 4.59% accuracy gain over standard RAG baselines. Notably, the 32B expertise-augmented LLMs pass all the computer-adaptive EMS certification simulation exams.
DeCoRL: Decoupling Reasoning Chains via Parallel Sub-Step Generation and Cascaded Reinforcement for Interpretable and Scalable RLHF
PDF ↗Existing reinforcement learning methods for Chain-of-Thought reasoning suffer from two critical limitations. First, they operate as monolithic black boxes that provide undifferentiated reward signals, obscuring individual step contributions and hindering error diagnosis. Second, sequential decoding has O(n) time complexity. This makes real-time deployment impractical for complex reasoning tasks. We present DeCoRL (Decoupled Reasoning Chains via Coordinated Reinforcement Learning), a novel framework that transforms reasoning from sequential processing into collaborative modular orchestration. DeCoRL trains lightweight specialized models to generate reasoning sub-steps concurrently, eliminating sequential bottlenecks through parallel processing. To enable precise error attribution, the framework designs modular reward functions that score each sub-step independently. Cascaded DRPO optimization then coordinates these rewards while preserving inter-step dependencies. Comprehensive evaluation demonstrates state-of-the-art results across RM-Bench, RMB, and RewardBench, outperforming existing methods including large-scale models. DeCoRL delivers 3.8 times faster inference while maintaining superior solution quality and offers a 22.7% improvement in interpretability through explicit reward attribution. These advancements, combined with a 72.4% reduction in energy consumption and a 68% increase in throughput, make real-time deployment of complex reasoning systems a reality.
S2D-Align: Shallow-to-Deep Auxiliary Learning for Anatomically-Grounded Radiology Report Generation
PDF ↗Radiology Report Generation (RRG) aims to automatically generate diagnostic reports from radiology images. To achieve this, existing methods have leveraged the powerful cross-modal generation capabilities of Multimodal Large Language Models (MLLMs), primarily focusing on optimizing cross-modal alignment between radiographs and reports through Supervised Fine-Tuning (SFT). However, by only performing instance-level alignment with the image-text pairs, the standard SFT paradigm fails to establish anatomically-grounded alignment, where the templated nature of reports often leads to sub-optimal generation quality. To address this, we propose S2D-Align, a novel SFT paradigm that establishes anatomically-grounded alignment by leveraging auxiliary signals of varying granularities. S2D-Align implements a shallow-to-deep strategy, progressively enriching the alignment process: it begins with the coarse radiograph-report pairing, then introduces reference reports for instance-level guidance, and ultimately utilizes key phrases to ground the generation in specific anatomical details. To bridge the different alignment stages, we introduce a memory-based adapter that empowers feature sharing, thereby integrating coarse and fine-grained guidance. For evaluation, we conduct experiments on the public MIMIC-CXR and IU X-Ray benchmarks, where S2D-Align achieves state-of-the-art performance compared to existing methods. Ablation studies validate the effectiveness of our multi-stage, auxiliary-guided approach, highlighting a promising direction for enhancing grounding capabilities in complex, multi-modal generation tasks.
I Have Covered All the Bases Here: Interpreting Reasoning Features in Large Language Models via Sparse Autoencoders
PDF ↗Recent LLMs like DeepSeek-R1 have demonstrated state-of-the-art performance by integrating deep thinking and complex reasoning during generation. However, the internal mechanisms behind these reasoning processes remain unexplored. We observe reasoning LLMs consistently use vocabulary associated with human reasoning processes. We hypothesize these words correspond to specific reasoning moments within the models' internal mechanisms. To test this hypothesis, we employ Sparse Autoencoders (SAEs), a technique for sparse decomposition of neural network activations into human-interpretable features. We introduce ReasonScore, an automatic metric to identify active SAE features during these reasoning moments. We perform manual and automatic interpretation of the features detected by our metric, and find those with activation patterns matching uncertainty, exploratory thinking, and reflection. Through steering experiments, we demonstrate that amplifying these features increases performance on reasoning-intensive benchmarks (+2.2%) while producing longer reasoning traces (+20.5%). Using the model diffing technique, we provide evidence that these features are present only in models with reasoning capabilities. Our work provides the first step towards a mechanistic understanding of reasoning in LLMs.
Harnessing the Unseen: The Hidden Influence of Intrinsic Knowledge in Long-Context Language Models
PDF ↗Recent advances in long-context language models (LCLMs), designed to handle extremely long contexts, primarily focus on utilizing external contextual information, often leaving the influence of language models' parametric knowledge underexplored. In this work, we firstly investigate how this parametric knowledge affects content generation and demonstrate that its impact becomes increasingly pronounced as context length extends. Furthermore, we show that the model’s ability to utilize parametric knowledge, which we call parametric recall ability, does not improve simultaneously with its ability to leverage contextual knowledge through extrinsic retrieval ability. Moreover, better extrinsic retrieval ability can interfere with the model’s parametric recall ability, limiting its full potential. To bridge this gap, we design a simple yet effective Hybrid Needle-in-a-Haystack test that evaluates models based on their capabilities across both abilities, rather than solely emphasizing extrinsic retrieval ability. Our experimental results reveal that Qwen-2.5 models significantly outperform Llama-3.1 models, demonstrating superior potential to combine various abilities. Moreover, even the more powerful Llama-3.1-70B-Instruct model fails to exhibit better performance, highlighting the importance of evaluating models from a dual-ability perspective.
In-context learning (ICL) enhances large language models (LLMs) by incorporating demonstration examples, yet its effectiveness heavily depends on the quality of selected examples. Current methods typically use text embeddings to measure semantic similarity, which often introduces bias in multi-step reasoning tasks. This occurs because text embeddings contain irrelevant semantic information and lack deeper reasoning structures. To address this, we propose GraphIC, a graph-based retrieval model that leverages reasoning-aware representation and specialized similarity metric for in-context example retrieval. GraphIC first constructs thought graphs—directed, node-attributed graphs that explicitly model reasoning steps and their dependencies—for candidate examples and queries. This approach filters out superficial semantics while preserving essential reasoning processes. Next, GraphIC retrieves examples using a novel similarity metric tailored for these graphs, capturing sequential reasoning patterns and asymmetry between examples. Comprehensive evaluations across mathematical reasoning, code generation, and logical reasoning tasks demonstrate that GraphIC outperforms 10 baseline methods. Our results highlight the importance of reasoning-aware retrieval in ICL, offering a robust solution for enhancing LLM performance in multi-step reasoning scenarios.
The Visual Prism: Refracting Images into Parallel Multilingual Descriptions with Structured Visual Guidance
PDF ↗Parallel corpora, as the foundation of machine translation, remain crucial even in the era of large language models (LLMs) for pre-training and fine-tuning. However, annotating parallel corpora is extremely costly, as it requires annotators to be proficient in multiple languages. To reduce this cost, prior work has explored image-pivoted corpus synthesis, generating multilingual captions for the same image as pseudo-parallel data. Unfortunately, these pseudo corpora suffer from the serious issue of multilingual focus divergence, i.e., the model attending to distinct aspects of the image when generating captions in different languages. To address this problem, we propose a method called PRISMS (Parallel Refracting ImageS into Multilingual descriptions with Structured visual guidance), which leverages semantic graphs as structured visual guidance to unify the focus of multilingual captions. To ensure adherence to this guidance, we introduce two key techniques: supervised fine-tuning using self-generated instructional data, and reinforcement learning with a reward signal based on semantic graph consistency. Experimental results on five languages show that our PRISMS significantly improves the image-pivot parallel corpora synthesis, enabling LLMs to achieve translation performance comparable to that of models trained on manually annotated corpora.
Simultaneous speech translation (SimulST) produces translations incrementally while processing partial speech input. Although large language models (LLMs) have shown strong capabilities in offline translation tasks, applying them to SimulST poses notable challenges. Existing LLM-based SimulST approaches either incur significant computational overhead due to repeated encoding of bidirectional speech encoder, or they depend on a fixed read/write policy, limiting the efficiency and performance. In this work, we introduce Efficient and Adaptive Simultaneous Speech Translation (EASiST) with fully unidirectional architecture, including both speech encoder and LLM. EASiST includes a multi-latency data curation strategy to generate semantically aligned SimulST training samples and redefines SimulST as an interleaved generation task with explicit read/write tokens. To facilitate adaptive inference, we incorporate a lightweight policy head that dynamically predicts read/write actions. Additionally, we employ a multi-stage training strategy to align speech-text modalities and optimize both translation and policy behavior. Experiments on both in-domain (MuST-C) and out-of-domain (Europarl-ST) En-De and En-Es datasets demonstrate that EASiST offers superior latency-quality trade-offs compared to several strong baselines.
Graph-of-Mark: Promote Spatial Reasoning in Multimodal Language Models with Graph-Based Visual Prompting
PDF ↗Recent advances in training-free visual prompting, such as Set-of-Mark, have emerged as a promising direction for enhancing the grounding capabilities of multimodal language models (MLMs). These techniques operate by partitioning the input image into object regions and annotating them with marks–predominantly boxes with numeric identifiers–before feeding the augmented image to the MLM. However, these approaches treat marked objects as isolated entities, failing to capture the relationships between them. On these premises, we propose Graph-of-Mark (GoM), the first pixel-level visual prompting technique that overlays scene graphs onto the input image for spatial reasoning tasks. We evaluate GoM across 3 open-source MLMs and 4 different datasets, conducting extensive ablations on drawn components and investigating the impact of auxiliary graph descriptions in the text prompt. Our results demonstrate that GoM consistently improves the zero-shot capability of MLMs in interpreting object positions and relative directions, improving base accuracy in visual question answering and localization up to 11 percentage points.
PASS: Probabilistic Agentic Supernet Sampling for Interpretable and Adaptive Chest X-Ray Reasoning
PDF ↗Existing tool-augmented agentic systems are limited in the real world by (i) black-box reasoning steps that undermine trust of decision-making and pose safety risks, (ii) poor multimodal integration, which is inherently critical for healthcare tasks, and (iii) rigid and computationally inefficient agentic pipelines. We introduce PASS (Probabilistic Agentic Supernet Sampling), the first multimodal framework to address these challenges for Chest X-Ray (CXR) reasoning. PASS adaptively samples agentic workflows over a multi-tool graph, yielding decision paths annotated with interpretable probabilities. Given the complex CXR reasoning task with multimodal medical data, PASS leverages its learned task-conditioned distribution over the agentic supernet. Thus, it adaptively selects the most suitable tool at each supernet layer, offering probability-annotated trajectories for post-hoc audits and directly enhancing medical AI safety. PASS also continuously compresses salient findings into an evolving personalized memory, while dynamically deciding whether to deepen its reasoning path or invoke an early exit for efficiency. To optimize a Pareto frontier balancing performance and cost, we design a novel three-stage training procedure, including expert knowledge warm-up, contrastive path-ranking, and cost-aware reinforcement learning. To facilitate rigorous evaluation, we introduce CAB-E, a comprehensive benchmark for multi-step, safety-critical, free-form reasoning. Experiments across various benchmarks validate that PASS significantly outperforms strong baselines in multiple metrics (e.g., accuracy, LLM-Judge, semantic similarity, etc.) while balancing computational costs, pushing a new paradigm shift towards interpretable, adaptive, and multimodal medical agentic systems.