Large Language Models (LLMs) often suffer from hallucination and struggle with complex reasoning tasks requiring multi-hop domain knowledge. While integrating Knowledge Graphs (KGs) provides a structured and verifiable information source, current KG-enhanced LLM paradigms usually rely on single-agent path extraction and fixed prompting, lacking adaptability and facing huge search spaces. To address these challenges, we propose RACER, a Reinforced Agent Collaboration framework for Explainable Reasoning on knowledge graphs. RACER employs a semantic-aware action pruning and teacher-guided reinforcement learning mechanism to efficiently extract high-quality reasoning pathways from large-scale KGs. Furthermore, to mitigate single-path generation pitfalls, we introduce a cross-task accumulated shared memory graph paired with an attention-driven multi-path knowledge refinement module. Finally, RACER orchestrates these components through a four-role multi-agent collaboration system (GraphAgent, TemplateAgent, AnswerAgent, and CriticAgent) to dynamically refine prompts and evaluate answers. Extensive experiments on CommonsenseQA and OpenBookQA datasets demonstrate that RACER significantly outperforms state-of-the-art KG-enhanced LLM baselines with an average improvement of 5\%, offering robust and highly interpretable reasoning capabilities.
Yuwei Lou, Hao Hu, Yuzhou Jiang et al.· 0 citations
Reinforcement fine-tuning (RFT) is increasingly used to strengthen the reasoning abilities of large models, yet its effectiveness is bound by how training data are selected and used. Most data-centric RFT methods rely on static or heuristic sample selection, implicitly assuming a sample's value is fixed over training. This overlooks the non-stationary dynamics of policy learning and can lead to suboptimal updates. We propose Dynamic Important Example Mining (DIEM), a principled and fully automated framework that makes data utilization adaptive throughout RFT. DIEM integrates two components into each optimization step: (i) a gradient-alignment importance estimator that efficiently approximates each sample's marginal contribution to policy improvement; and (ii) a constrained batch reweighting scheme that maximizes aggregate utility while preserving the update's gradient magnitude to stabilize optimization. Across several reasoning benchmarks, DIEM consistently outperforms strong static and dynamic baselines. The code will be released via https://github.com/hrtan/DIEM.
Haoru Tan, Sitong Wu, Yanfeng Chen et al.· 0 citations
Failures in agentic AI systems can arise from interactions among messages exchanged by multiple large language model (LLM) agents. Pointwise attribution cannot distinguish a jointly necessary repair from alternative singleton repairs. We formulate Minimal Repair Family Recovery (MRFR): recovering all inclusion-minimal event sets whose counterfactual replay restores task success within a declared size bound. We propose Graph-Constrained Joint Replay (GCJR), which slices failure-relevant events from an execution dependency graph, constructs graph-feasible singleton and pair candidates, and verifies them by replay with paired clean counterparts. For fixed replay outcomes, GCJR is exact within its declared graph domain. On 90 in-scope cases from a 120-DAG controlled benchmark, GCJR achieves 1.000 Family Exact Match while reducing mean replay calls from 56.3 to 25.3 (55.1%) relative to exhaustive search. On a 24-case, four-agent LLM pilot, it again achieves 1.000 Family Exact Match and reduces mean model calls from 21.0 to 10.0 (52.4%); single-event replay misses jointly necessary repairs.
Bingjie Li, Yumeng Song, Zhongming Yao et al.· 0 citations
Accurate through-thickness measurement of subsurface delamination depth in Carbon Fiber Reinforced Polymer (CFRP) is important for structural assessment because defect location determines affected load-bearing layers. Optical pulsed thermography (OPT) provides a two-dimensional thermal video rather than volumetric measurements, so depth must be inferred from temporal heat-diffusion responses. A challenge is spatial dataset bias: when calibration defects follow regular grids, regression models may memorize their geometry instead of learning physical relationship between thermal decay and depth. This work introduces a spatio-temporal decoupling architecture that separates spatial defect localization from temporal depth measurement. Defect regions are first localized using segmentation methods, after which thermal responses are spatially averaged and converted into sixteen physics-informed temporal, energy, statistical, and geometric features. These features expose the one-dimensional heat-conduction relationship while withholding pixel coordinates from the depth model. Four regression models are evaluated using specimen-level cross-validation: Random Forest (RF), Gradient Boosting Machine (GBM), Advanced Multi-Layer Perceptron (Adv-MLP), and XGBoost. Unregularized trees and over-parameterized Adv-MLP exhibit calibration collapse under geometric shifts, with errors exceeding 0.5 mm. In contrast, regularized XGBoost with L1/L2 penalties and column sampling maintains cross-specimen calibration, achieving a mean absolute error (MAE) of 0.056 mm and root mean square error (RMSE) of 0.085 mm. Predicted depths are merged with masks to generate Delaunay-triangulated three-dimensional defect models in three to five seconds per specimen. Results show that mathematical regularization and spatio-temporal decoupling reduce spatial memorization in thermal-video depth regression.
Zain Ul Abidin, Habeeban Memon, Junaid Ahmed· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Text-to-Image (T2I) models have recently achieved impressive visual fidelity, yet their evaluation remains constrained by benchmarks that are often difficult to interpret and insufficiently diagnostic. Existing skill-based evaluations tend to overlook critical failure modes that strongly impact usability but fall outside standard taxonomies, such as global incoherence arising from missing parts or physically implausible configurations (e.g., floating objects). In addition, prompt difficulty is typically controlled along a single dimension; either prompt length or the number of elements to generate. To address these limitations, we introduce Imag-Eval, a controlled benchmark designed to assess how T2I models ground compositional natural-language instructions into visual outputs. Unlike prior work that conflates surface linguistic complexity with compositional difficulty, Imag-Eval explicitly seeks to disentangles these factors by independently varying both the number of instances and the combination of constraints (rules), while avoiding error propagation. This design enables fine-grained and interpretable analysis of where cross-modal instruction following fails. Our benchmark comprises 1,140 prompts and 8,842 combined rules, and we evaluate it on several state-of-the-art models. Complementing this analysis with an additional study of over 2,000 prompts from a concurrent benchmark, our results suggest that, for structured skills, compositional difficulty is primarily governed by the number of grounded rules and their binding to instances,, rather than by prompt length alone.
Ibrahim Mohamed Serouis, David Jaramillo Duque· 0 citations
Bias in human-agent interaction can manifest not only through hostile language but also as benevolent bias, whereby unequal treatment hides behind a warm, positive tone. To make it detectable, we operationalise benevolent bias along two dimensions, tone and treatment, yielding three classes: neutral support, overt bias, and benevolent bias. Building on these definitions, we construct BENEVDIAL, a class-balanced corpus of 362,880 multi-turn support dialogues spanning user and agent demographics, roles, and generators, to support controlled evaluation. We then test two detector families on it: off-the-shelf safety detectors and prompted large language model (LLM) judges. Our results reveal a detection gap: off-the-shelf detectors reliably flag overt bias yet largely miss benevolent bias, while LLM judges catch more under more explicit detection criteria but increasingly misclassify neutral support as benevolent bias, and demographic context amplifies the false alarms. These findings suggest that fair monitoring of human-agent dialogue must look beyond surface cues to whether the agent's treatment is disparate.
Qianqi Liu, Jin Huang, Fethiye Irmak Dogan et al.· 0 citations
Temporal inconsistencies, such as mandates attributed outside their real interval, events presented as past before they occurred, or inverted causal sequences, are a form of political disinformation that evades style-based fake news detectors: a well-written article with a single wrong date carries no lexical signal of falsehood. This paper introduces the Temporal Coherence Score (TCS), a continuous, intrinsically interpretable metric that quantifies the temporal coherence of a news article, computed by a four-stage pipeline: extraction of temporal facts, construction of a temporal knowledge graph, hierarchical verification against internal consistency rules and external reference sources, and score aggregation with automatically generated explanations. Verification combines eight internal checkers derived from Allen's interval algebra with a five-level external hierarchy ranging from a locally stored reference knowledge base of 1{,}256 curated political facts to live Wikidata SPARQL queries. On a benchmark of 100 political news articles with injected temporal errors, the system reaches a precision of 0.909 at the selected operating threshold, with a single residual false positive, a profile deliberately tuned for human-in-the-loop fact-checking assistance, where false alarms are costlier than missed detections. Unlike lexical baselines that output only a binary label, every flagged article is accompanied by the inconsistency type, the entities involved, and the reference source that contradicts the claim.
The LLM-as-a-Judge paradigm has emerged as a scalable alternative to human evaluation. However, single-model judges are limited by their inherent model biases, while multi-agent evaluation protocols that mitigate this through diverse deliberation are prohibitively expensive at inference time. To this end, we propose \textbf{\modelname}, which equips a compact \underline{Judge} model with multi-agent \underline{Panel} deliberation capability. Specifically, we first train on panel deliberation traces from an ensemble of strong evaluators, capturing structured patterns of discussion, disagreement, and resolution. To further improve judgment quality beyond SFT, we introduce \textit{AdaReward}, an adaptive multi-reward RL algorithm that dynamically rebalances reward component weights as different objectives saturate at different rates during RL training. For practical deployment, we further design a lightweight domain specialization module for rapid adaptation to new evaluation domains with few hundred labeled samples. As a result, (i) \textit{Novel}: the first framework to equip a single compact judge with multi-agent panel deliberation capability at single-model inference cost; (ii) \textit{Effective \& Reliable}: JudgePanel with a 14B backbone outperforms judge-specialized models up to 70B across four evaluation benchmarks, demonstrates strong position consistency, and rapidly specializes to new domains with few hundred samples.
Yiyue Qian, Shinan Zhang, Huan Song et al.· 0 citations
Learning effective multimodal entity representations is fundamental for reasoning tasks such as multimodal knowledge graph completion (MMKGC). However, existing methods often suffer from semantic over-smoothing within modalities and ineffective noise filtration across modalities, particularly under sparse or ambiguous conditions. To overcome these limitations, we propose PrismF, a unified framework that synergizes multi-perspective enhancement with progressive fusion to extract stronger signals from diverse inputs. PrismF enhances fine-grained intra-modal semantics through a multi-perspective mechanism that decomposes each modality into complementary views and constrains them with a decoupling loss to reduce representation collapse. Furthermore, it improves cross-modal integration through a progressive fusion strategy that dynamically calibrates inter-modal interactions, enabling the model to emphasize informative signals while suppressing noisy or unreliable ones. Extensive experiments on three public benchmarks show that PrismF achieves the strongest overall performance, including relative improvements of 4.04% in MRR and 11.17% in Hits@1 on KVC16K. Our code can be found at https://github.com/HubuKG/PrismF.
Chenyi Xiong, Yan Zhang, Jing Hu et al.· 0 citations
We propose a scalable, validity-oriented pipeline for evaluating biomedical LLM judges when high-quality human judgments are scarce. First, we augment existing human-labelled biomedical benchmarks with deterministic, metric-grounded mutations that produce auditable preference pairs. Second, we evaluate judges beyond aggregate correctness using three deployment-relevant dimensions: correctness against metric-derived gold labels, robustness under repeated stochastic sampling, and compliance with the requested output format. We use this pipeline to assess Llama-3.1-8B-Instruct under four regimes: (1) base, using the instruct model as is; (2) SFT, distillation-based supervised fine-tuning only; (3) RL, GRPO-based reinforcement learning only; and (4) SFT$\rightarrow$RL, SFT followed by RL. The base and single-stage regimes struggle on structured medical discrimination such as PICO extraction and clinical calculations, whereas SFT$\rightarrow$RL performs best across correctness, compliance, and robustness; gains concentrate on decomposable tasks (PICO, MedCalc), at times matching or outperforming frontier models.
Rodrigo de Oliveira, Federico Pittino, James Gwinnutt et al.· 0 citations
Multimodal safety moderation requires distinguishing risks arising from visual content, user intent, and assistant behavior. Existing safeguards, however, are typically trained for a single judgment target and reduce safety assessment to a binary decision. Consequently, risk becomes difficult to compare across a multimodal interaction, and ambiguous cases are obscured. We introduce SafeAtlas-VL, a dataset of 1.5M training instances that places image-, request-, and response-level judgments on a five-level ordered scale. We curate a broad collection of safety-relevant data from both real-world and synthetic sources and apply a disagreement-aware annotation procedure. The resulting dataset spans 15 harm categories and 55 fine-grained subcategories, covering a broad range of multimodal safety scenarios. We also construct SafeAtlas-Bench, a held-out set of 5,000 instances for evaluating five-level predictions and continuous risk scores. Upon this dataset, we train the SafeAtlas Guard series of models via target-conditioned tuning for multimodal safety detection. Our models not only perform five-way classification of safety levels but also map safety to continuous scores through a soft cumulative ordinal head. Experimental results demonstrate that guard models trained on our dataset exhibit strong generalization: even without using the training sets of other benchmarks, they achieve competitive performance on the corresponding test sets. Notably, our 8B model attains the overall best performance, outperforming the previous SOTA by approximately 4% in F1 score. Code, data, and models are released to support further research. Warning: this paper contains example data that may be offensive, harmful, graphic, or disturbing.
Zongrui Wang, Xiangyang Zhu, Sicheng Wang et al.· 0 citations
Large vision-language models (LVLMs) frequently generate content unsupported by visual inputs. Preliminary experiments show that visual evidence is primarily incorporated into answer-side representations in early-to-middle decoder layers, while its direct influence progressively weakens in later layers. This attenuation suggests that visual evidence acquired earlier may be insufficiently utilized during subsequent generation. Based on this observation, we propose EviAnchor, a training-free and single-branch inference framework that preserves and reactivates visual evidence throughout generation. EviAnchor introduces Regional Evidence Anchor (REA) slots to progressively aggregate dense visual tokens into spatially structured representations. It then strengthens the current decision state's access to these visual anchors through decision-conditioned evidence routing, mitigating excessive dependence on textual context. Finally, the model resumes its native Transformer computation to integrate the retrieved visual evidence with question semantics and generation history. Experiments across POPE, CHAIR, and MMHal-Bench demonstrate consistent improvements in visual grounding.
Sihang Jia, Shuliang Liu, Songbo Yang et al.· 0 citations
