每天知道
你的领域
出了什么

Multimodal expansion of large language models (LLMs) enables new perceptual capabilities but often compromises the language intelligence acquired during pretraining. In this work, we investigate this phenomenon from the perspective of internal adaptation dynamics and discover that neurons in pretrained LLMs exhibit heterogeneous plasticity during multimodal learning: some neurons are critical for preserving language capabilities, while others are more adaptive to multimodal knowledge. Based on this insight, we propose NeuPAT (Neuron-aware Plasticity Allocation Tuning), a lightweight and architecture-agnostic framework that allocates neuron-wise update constraints during multimodal instruction tuning. NeuPAT uses a small-scale probing stage to estimate neuron adaptation patterns and selectively protects language-sensitive neurons while promoting multimodal adaptation through more plastic neurons. Experiments across diverse LLM families demonstrate that NeuPAT recovers 94.5% of the language capability degradation caused by vanilla tuning on 11 language benchmarks while maintaining comparable multimodal performance, providing an effective approach for capability-preserving multimodal expansion.

Test-time compute scaling is a primary driver of performance in large reasoning models (LRMs), but extreme inefficiency bounds current approaches, shifting the critical question from how much compute to spend, to where to allocate it. We formalize test-time reasoning as a constrained compute allocation problem over partial trajectories. Under a fixed hardware budget, existing paradigms fail to actively allocate the compute to the most promising partial progress: traditional parallel sampling treats traces independently and induces severe memory bottlenecks, while subtractive pruning starves hardware and fails to actively and sufficiently shift the output distribution. To overcome this dichotomy, we introduce Gambit, an inference algorithm that executes thought-level beam search. By periodically pruning unpromising trajectories and immediately branching from high-quality prefixes, Gambit dynamically concentrates compute onto the most promising reasoning traces via a light-weight scorer probing hidden states while maintaining continuous high hardware utilization. Extensive evaluations across multiple models and benchmarks demonstrate that Gambit strictly dominates existing baselines. Under identical hardware constraints, our method yields up to a +6.7% absolute accuracy gain on HMMT-24 and +3.3% on AIME-25 over pruning baselines, delivers $>2\times$ higher throughput on trace completion, and reduces total token consumption by up to 68.5% relative to standard parallel sampling.

Large language models (LLMs) are increasingly reported to exhibit human-like neural and cognitive signatures, including concept cells, mental number lines, and cognitive maps. These claims often rely on linear probing and activation steering applied to a single model, yet both methods are highly sensitive to measurement choices. A reported parallel may therefore reflect the model, the measurement procedure, or both. We audit four representative neuroscience-inspired paradigms across 17 models from five families, spanning $0.6$B to $72$B parameters. Our main experiment examines the causal steerability of concept directions. With raw activation units and a fixed layer and coefficient, steerability appears to increase with model scale, resembling an emergent capability. However, this pattern is produced by an uncalibrated pipeline rather than by a claim established in the steering literature. The trend depends jointly on raw units, the readout metric, and the operating point; correcting any one of these removes it. With residual-norm-comparable interventions and held-out operating-point selection, concept steering remains significant at every scale, but shows no significant trend across the Qwen3 series, although the confidence interval does not rule out a moderate positive slope. The remaining results are mixed. A linear geographic world map is consistently decodable in every tested checkpoint up to $72$B. Number magnitude is strongly encoded, but whether individual neurons appear bell-shaped or monotonic depends on the selection criterion. Language-specific structure is localizable, but the direction of the cross-lingual asymmetry reverses under a different attribution method. These results suggest that the main constraint on AI neuroscience is not a lack of phenomena, but a lack of comparable measurements and adequate controls. We release the protocol, stimuli, and code.

In-context learning (ICL) can induce emergent misalignment (EM), where narrow misaligned examples alter answers to unrelated questions. Existing prompts, however, conflate harmful-text exposure with an invitation to continue assistant behavior. We hold harmful answers fixed while varying their delivery as demonstrations, evidence, assistant history, or tool output. Across ten independently sampled contexts, demonstration framing raises broad EM by $30$--$32$ percentage points on a susceptible Gemini model; the gap survives domain exclusion, semantic clustering, unseen questions, and four prompt templates. Format and length-matched controls show that harmful content is necessary but insufficient. A role times continuation factorial further reveals model-dependent provenance effects: Gemini follows both assistant and tool histories, whereas Grok largely resists tool-framed continuation. Several other frontier and open-weight models show no gap. Blinded human audits confirm every main contrast and show that the model judge underestimates active-condition failures. Thus continuation framing is a strong, model-dependent moderator of ICL-EM, not a universal consequence of harmful context.

Sports video analysis is crucial for athletic analytics and broadcasting enhancement. Dense sports video reasoning, however, demands a fine-grained understanding of numerous small-scale, highly interactive, and visually homogeneous entities (e.g., players sharing identical uniforms, the ball) across long temporal contexts. Current Large Multimodal Models (LMMs) inherently struggle with such dense visual complexities. Due to the lack of fine-grained visual details, these models often over-rely on textual priors to guess answers, especially when distinguishing visually similar actions and players. To address this, we propose SportsGrounder, a framework that leverages an open-vocabulary visual expert to aid interleaved grounding specifically for dense sports video reasoning. To achieve precise spatial localization, we extract domain-guided object proposals and introduce an Interleaved Grounding Fusion (IGF) mechanism. The IGF frame-by-frame integrates explicit bounding box coordinates and implicit visual semantics with global grid features. This design preserves strict temporal alignment and prevents sequence length explosion. Furthermore, we design an Action-Aware Supervision (AAS) module that directly regularizes the model's hidden states, forcing the network to learn accurate motion representations rather than relying on language bias. Optimized with Mixed Preference Optimization (MPO) to better distinguish deceptive distractors, our extensive experiments on newly curated dense sports VQA datasets (derived from SoccerNet and FineSports) demonstrate that SportsGrounder significantly improves fine-grained reasoning and achieves state-of-the-art accuracy.

Large language model (LLM) fine-tuning at the edge adapts the model to scenario-specific data while preserving privacy. Although existing studies proposed pipeline parallelism to address the limited memory and computing resources of edge devices, they commonly rely on backpropagation (BP) training, which has a fundamental limitation of update locking and could experience severe throughput and memory bottlenecks. In this work, we propose a BP-free algorithm, called ZeroLock, that decouples the model updates into independent chunk updates by local objective construction. It breaks the update locking of BP and hence can improve throughput at the algorithm level and lower memory usage by reducing activation storage. To the best of our knowledge, we provide the first theoretical framework for such local objective construction-based approaches under general model chunk division by mapping local objectives to the global objective. We prove that ZeroLock has a convergence rate of $\tilde{\mathcal{O}}(1/\sqrt{T})$, which differs from BP only by polylogarithmic factors. We design a system for ZeroLock and build real-world prototypes, incorporating techniques such as early forwarding and failure recovery for efficient and robust implementation. Experiments on the prototype show that compared to BP-based baselines, ZeroLock reduces the memory by 26.5% and improves throughput by 4.9%.

Residual connections rely on a static residual pathway, and are essential for training deep neural networks. Hyper-connections (HC) increase the expressivity of residual routing by incorporating multiple residual streams and learning dynamic information flow, while manifold-constrained (mHC) variants stabilize training through doubly stochastic residual mixing. However, a generator-level bottleneck remains in existing methods: they use dense, unstructured generators for pre-branch aggregation, residual mixing, and post-branch redistribution, which results in parameter count growing rapidly with the number of streams. To address this issue, we propose \underline{T}ensorized \underline{E}fficient \underline{M}anifold-constrained \underline{P}arameterization for \underline{E}xpressive Residual \underline{R}outing (TEMPER), which represents these generators as multi-way tensors over the input-stream, feature, and output-stream modes, and parameterizes them using tensor networks. Such a structured low-rank formulation is shown to preserve token-dependent manifold-constrained routing interface while substantially reducing parameter growth. It also promotes interpretability and intuition, as: i) tensor ranks control the dimensionality of the learned routing subspace, with full ranks recovering dense routing; while ii) the generator approximation errors bound differences in routing logits and, consequently, in the routed-block outputs. Comprehensive experiments show that TEMPER matches or outperforms existing methods across language modeling and commonsense reasoning tasks, while requiring substantially fewer additional parameters. At eight residual streams, TEMPER achieves the best CORE score while using about $84%$ fewer additional parameters than mHC, thus showing a stronger performance-parameter efficiency trade-off.

Reasoning language models increasingly use test-time compute to improve performance, but existing evaluations typically study this compute one question at a time. Yet when multiple problems share an end-to-end cost or latency constraint, models must decide how to divide limited inference compute among them. We introduce an exam-style evaluation framework for studying this setting, in which a model must distribute one shared token budget across questions with different difficulty and point values to maximize its total score. Across several open and frontier reasoning models, we find that models fail to allocate a shared budget strategically across questions of varying difficulties and values. Models behave largely as greedy sequential solvers: they prioritize questions by presentation order, front-load effort on early questions, and remain insensitive to value, with these tendencies becoming more pronounced as the number of questions grows. Explicit planning prompts spread compute more evenly but do not produce value- or difficulty-aware prioritization. The same behavioral pattern extends from mathematical to code reasoning. These findings establish global budget allocation as a distinct capability that is not captured by conventional per-question evaluation and remains a challenge for current reasoning models.

Large language models (LLMs) are increasingly involved in group decision-making with other LLMs and humans. Yet it remains unclear whether their influence is driven by persuasion-oriented expression or compliance-oriented accommodation. We introduce DecisionQE, a questionnaire-based framework for measuring each model's persuasive and compliant tendencies across multiple decision scenarios, and use the Werewolf game as an interactive testbed to study their effects on social influence and group outcomes under asymmetric information. Across experiments, stronger persuasive tendency does not significantly improve group outcomes, whereas compliant-oriented models show more stable advantages in cooperation. We further reveal a dual effect of compliance: it supports cooperation in honest roles but improves concealment in adversarial roles. These findings suggest that LLM group interactions reveal not only task outcomes, but also measurable patterns of intrinsic behavioral tendency. LLMs can therefore serve as a lens for sociological observation of language-mediated interaction, while highlighting the need to incorporate behavioral tendencies into safety evaluation of LLM systems.

Large language models (LLMs) are high-value assets that can be derived through redeployment, fine-tuning, quantization, or further alignment. Because deployed LLMs are commonly exposed only through query APIs, ownership verification must often rely on black-box text responses. This setting is difficult: generations are open-ended and can vary across repeated queries, while existing black-box fingerprints rely on signals that are fragile under a final-response interface, including full-text matching, soft behavioral features, or model-specific prompts designed not to transfer. We propose TCF (Targeted Counterfactual Fingerprinting), a black-box LLM fingerprinting framework that converts open-ended generation comparison into constrained-answer targeted counterfactual transfer. TCF restricts each verification query to a finite answer space, reducing the surface-form ambiguity that enters the verification score, and optimizes a prompt perturbation toward a counterfactual target different from the protected model's clean answer on the original prompt. Verification reduces to checking whether the suspect model's parsed final answer matches the recorded target. We introduce the source-model counterfactual margin (SCM), a protected-model-only quantity that certifies the target is unlikely before the perturbation and likely after it; SCM controls target selection, perturbation stopping, and fingerprint filtering. Under explicit derived-preservation and independent-transfer budgets motivated by local behavioral closeness, we derive a target-accuracy gap between derived and independent models. Across four LLM families, TCF achieves an average AUC of 0.9861, improving over TRAP, ProFLingo, and ZeroPrint by 0.07 to 0.19.

Recent advances have highlighted the potential of machine learning, particularly Large Language Models (LLMs), for analyzing and optimizing programs. We present the first application of program embeddings from LLMCompiler---an LLM massively pretrained on intermediate representation (IR) code---to representative program analysis and optimization tasks. We generate program embeddings directly from source and IR code using a simple approach: split programs into chunks, independently embed each chunk with pretrained LLMs, and then aggregate the chunk embeddings into a single program embedding. Our experiments show that combining source and IR code embeddings achieves an error rate of 1.54% in algorithm classification, a 12% improvement over the current state-of-the-art, and a competitive accuracy on heterogeneous device mapping. These findings suggest that training a performance-aware LLM for embedding IR code might yield state-of-the-art results in code optimization tasks.

We extend the DS@GT ARC working-note submission to the Touché 2025 Retrieval-Augmented Debate task. The task has two subtasks: generating the next utterance in a simulated debate, and evaluating debate responses according to the Gricean maxims of Quantity, Quality, Relation, and Manner. The DS@GT ARC submission consisted of six leading LLMs from three providers through a retrieval-augmented prompting pipeline. We summarize the results from the working paper and explore whether multi-LLM evaluator agreement is a reliable proxy for official evaluation performance. The analysis shows that frontier LLM systems are strong response generators, and as evaluators they agree strongly within model families. However this consensus does not reliably track the official evaluation target, with the largest gap on the Quality maxim. The accompanying source code for this paper is located at https://github.com/dsgt-arc/touche-2025-rad and https://github.com/dsgt-arc/touche-2025-rad-analysis.

Fake news increasingly relies on cross-modal image-text forgeries, making transparent and verifiable reasoning chains an urgent need for Detecting and Grounding Multi-Modal Media Manipulation (DGM4). Existing methods produce black-box detection results without any decision rationale, limiting their reliability in forensic practice. Multi-modal Large Language Models (MLLMs) offer a natural path toward explainability, but applying them to DGM4 raises two difficulties. First, models tend to generate explanations disconnected from predicted evidence locations, producing unverified attribution. Second, enforcing evidence-conclusion consistency requires active optimization, yet uniform training signals fail to distinguish localization tokens from classification tokens, making multi-head joint training unreliable. We propose a multi-modal manipulation detector based on an Evidence-Grounded Forensic Reasoning (EFR) framework. EFR introduces an Anchor-and-Verify reasoning chain that enforces modality-isolated perception before cross-modal comparison, with conclusion coordinates as explicit anchors to which downstream evidence must spatially correspond. A verifiable reward system then enforces evidence-conclusion consistency during training, while a Modality-Decoupled Advantage (MDA) routing mechanism mitigats credit misassignment across prediction tasks. Experiments show that EFR achieves state-of-the-art performance while producing structured forensic reasoning records that explicitly bind explanations to evidence.

While Large Language Models (LLMs) employing Chain-of-Thought (CoT) exhibit superior reasoning capabilities, the neural mechanisms distinguishing this explicit Thinking mode from direct answer generation (NoThinking mode) remain poorly understood. To deconstruct this cognitive process, we apply Top-K Sparse Autoencoders (SAEs) to the intermediate representations of DeepSeek-R1-Distill-Qwen-7B and examine the model's divergent behaviors across math-solving tasks of three distinct difficulty levels. Observationally, we identify a clear distinction in how the model functions under two reasoning modes: Thinking mode relies on sparse and high-intensity feature activations driving verbal deduction independent of problem complexity, whereas NoThinking mode exhibits an adaptive and diffuse pattern prioritizing symbolic manipulation. Causally, suppressing the three most active sparse features by Total Activation Volume reveals three principles: (i) reasoning and syntactic structure are tightly coupled, as interventions consistently degrade \LaTeX{} and boxed-solution formatting; (ii) Thinking responds to disruption with compensatory over-generation marked by increased metacognitive cues and repetitive, low-information continuations; and (iii) coherent CoT behavior depends on a fragile coordination among specialized features, yielding distinct failure modes under perturbation but a consistently impaired output structure.

Video Anomaly Detection (VAD) aims to identify anomalous events and localize their temporal intervals. Existing approaches exhibit a "when-what" dissociation: traditional DNN-based methods localize when anomalies occur but lack semantic understanding, whereas LLM-based methods explain what happens but neglect precise temporal grounding. We attribute this to the absence of a unified reasoning paradigm. Inspired by how humans inspect surveillance videos - glancing globally to form temporal hypotheses, scrutinizing suspicious segments, and thinking iteratively to correct errors - we study this global-to-local paradigm from two perspectives. We first propose Glance then Scrutinize (GtS), a training-free framework using static and dynamic textual guidance for coarse-to-fine anomaly grounding and understanding, balancing accuracy and speed. To break the ceiling imposed by frozen external modules, we further propose a tool-augmented agentic VAD method, where a multimodal large language model learns to invoke a video cropping tool, inspect densely resampled frames, and self-correct mislocalized hypotheses, via cold-start supervised fine-tuning followed by reinforcement learning with a joint answer-grounding reward. For training and evaluation, we extend our prior VAGU benchmark into VAGU-T (Video Anomaly Grounding, Understanding, and Thinking), comprising 7,567 real-world videos over 21 anomaly categories with human-validated grounding, explanations, QA pairs, and chain-of-thought tool-calling traces. We further introduce JeAUG, a metric jointly evaluating semantic interpretability and temporal precision. Experiments show that GtS substantially surpasses training-free baselines, while the agentic model delivers both higher accuracy and faster inference.

Transformers have had a profound impact on the world of language processing and computer vision. As efforts to answer the million-dollar question of ``How does a Transformer learn?" have been increasing, existing interpretability studies primarily analyze representations at isolated layers or the network as a whole, while the developmental evolution of individual representations and its manifolds across transformer layers remains underexplored. With this work, we aim at providing a comprehensive analysis of the evolution of representations as the representation point cloud transforms across the layers; thereby attempting to isolate layers or establish a trend which comes closer to justifying how and when raw input representations evolve into task-relevant feature representations. Thus, Transformer Geometry Observatory-TGO-IV introduces a topological framework for analysing the evolution of Transformer representations through the lens of Persistent Homology. Rather than studying local geometric properties alone, TGO-IV constructs Vietoris--Rips simplicial complexes from token-level representation point clouds and investigates the evolution of their persistent topological signatures across Transformer layers. The proposed framework comprises complementary topological observatories including Persistence Diagrams, Barcode Diagrams, Betti Curves, Persistence Landscapes, Bottleneck Distance, and Wasserstein Distance, enabling a comprehensive analysis of how the global topology of representation point clouds develops throughout the forward pass.

Image-based profiling captures rich phenotypic signatures for drug discovery and functional genomics. Large public datasets like JUMP Cell Painting now provide millions of images for systematic study. However, the scale of these resources, 115 TB for JUMP alone, and fragmented evaluation practices make systematic comparison of representation methods intractable for many researchers. Here we present Nahual, an open-source framework for reproducible model deployment, and JUMP-lite, a curated 116 GB subset of JUMP that is 1000 times smaller while preserving phenotypic diversity through careful selection of perturbations with high-confidence annotations and a storage reduction via lossy JPEG XL compression. With these, we benchmark five representation methods, including classical features (CellProfiler) and deep learning models (MorphEM, OpenPhenom, SubCell, DINOv2), and demonstrate that compression preserves downstream signal while standardized phenotypic activity and consistency metrics reveal meaningful performance differences across methods. Together, JUMP-lite and Nahual provide a foundation for accessible, reproducible benchmarking of image-based cell representations.

Traffic data collection is dominated today by deep object detectors followed by tracking-by-detection, a pipeline that presupposes what is often missing in practice: a detector already trained on the class one wants to count. We revisit a purely geometric traffic-sensing pipeline for Single Board Computers in which detection is class-agnostic: moving objects come from background subtraction and thresholding, and counting is decided by a geometric rule on an imaginary line across the road, a software inductive loop detector. With no object model, training set or per-object trajectory, it runs faster than real time on Raspberry Pi class hardware. Two counting rules are described: a constant average speed rule, whose expected accuracy is derived analytically as about 86% under a Gaussian speed distribution, and a self-calibrating pre-calibration rule that recovers the lane geometry from blob statistics and counts edges of lane occupancy, additionally yielding per-vehicle average speed at no extra cost. Over four videos the latter counts with 83.3%-100% accuracy; in a field deployment it reaches 91% against 37.5% for a blob-tracking baseline under the same compute budget. We report the observations of that period in detail: the resolution floor below which accuracy collapses, the frame rate floor at which vehicles alias past the counting line, the gap between short curated clips and long uncontrolled footage, and the trade-off between Python (easier to tune, 100% CPU) and C++ (40% CPU, thermally viable). These are properties of the sampling geometry, not of the hardware of the time, and still constrain edge deployments. We close by arguing where motion-based, class-agnostic detection remains the right tool: open-set classes with no annotated data, tight power budgets, privacy-constrained installations, and the cold start of mining training crops to bootstrap a learned detector.

Traditional block-based video codecs, such as H.264/AVC, H.265/HEVC and H.266/VVC, rely on hand-crafted Rate-Distortion Optimization (RDO) processes that primarily minimize Mean Squared Error (MSE), which correlates poorly with human perceptual quality. While neural video compression methods can easily optimize perceptually aligned metrics like MS-SSIM, their high computational complexity limits practical deployment. This paper proposes a novel bit allocation transfer framework that bridges these two paradigms to enhance the perceptual quality of conventional video codecs. Specifically, we train a quantization step generation model using a perceptual loss within a neural video compression framework (DCVC-FM). The model takes the original frame and a motion-compensated prediction as input and outputs a quantization step map. We then derive a block-wise bit ratio from this map and convert it into a Quantization Parameter (QP) map for a traditional video codec. Experimental results on the HEVC B$\sim$D dataset demonstrate that our method achieves 20.20%, 8.25%, and 8.37% bitrate savings in terms of MS-SSIM compared with the standard reference software JM-19.0, HM-16.20, and VTM-23.0, respectively, with additional gains when utilizing predicted frames. Our approach effectively transfers the implicit perceptual importance learned by neural video compression models to guide block-level bit allocation in traditional video codecs without modifying their core decoding syntax.

Vision-Language-Action (VLA) models achieve strong robotic manipulation performance but often degrade under visual and environmental shifts. Latent world modeling offers a promising approach to improving robustness, yet existing methods commonly encode camera views independently and predict holistic scene dynamics without explicitly modeling their geometric relationships. We propose GWM-VLA, a geometry-aware latent world modeling framework for VLA learning. GWM-VLA combines geometry-aware multi-view state encoding, global context-conditioned target-view prediction, and shared latent-action representations grounded by robot-action supervision. Specifically, VGGT-$Ω$ jointly aggregates multi-view observations at each timestep to construct geometry-aware multi-view states. The latent world model predicts the next-step patch tokens of a selected target view using patch and register tokens obtained after multi-view aggregation, thereby retaining multi-view geometric information without predicting the complete multi-view state. We use the wrist view as the target in our experiments, placing greater emphasis on end-effector motion and local gripper-object interactions. Finally, the shared latent-action representations condition both the latent world model and the flow-matching action head, allowing latent-prediction supervision and ground-truth robot-action supervision to jointly shape the same latent-action representations. Experiments across both simulation and real-world environments demonstrate the effectiveness and robustness of GWM-VLA.

A world model is only useful for physical AI if it changes what the agent does, and only safe if it declines to do so when it is wrong. We study both halves of that requirement with CausalNav, a controller built around a signed, action-conditioned transition graph over identified state coordinates. At deployment CausalNav simulates a small library of intervention sequences, converts their objective error into policy-logit advice, and admits that advice only when a scale-free predictive-reliability certificate, a policy-margin gate, and an argmax-agreement gate all pass; otherwise it falls back exactly to its own model-based base controller. We evaluate against nine controlled baselines (transformer, recurrent, split-latent, graph, causal-induction, and three recent model-based reasoning modules) on CartPole-v1 and discretized Pendulum-v1 with physical-parameter shifts, under one shared PPO trainer, one interaction budget, and ten held-out seeds (200 runs). CausalNav attains the best average rank (1.25 of ten). The diagnostic result is more informative than the ranking: the learned graph recovers structure well above chance (CartPole F1 = 0.59 +/- 0.09), yet per-seed structural fidelity is uncorrelated with per-seed control benefit (r = -0.15, p = 0.67), and the certificate abstains on 10/10 Pendulum seeds, where forcing the planner on costs return. Model fidelity did not predict downstream control utility in our setting; certified abstention, not better prediction, is what made the world model safe to deploy.

Recent advances in flow matching models have substantially improved the quality of text-to-image generation, but have also raised increasing safety concerns due to their potential to generate harmful or undesirable content. Existing concept erasure methods for flow matching models predominantly focus on removing individual concepts, while effectively erasing multiple concepts simultaneously remains challenging. We propose FlowErase-OPD, a framework for multi-concept erasure based on on-policy distillation (OPD). Our approach first distills multiple single-concept erased models into a unified LoRA module and introduces Anchored Multi-Teacher Distillation (AMTD), which incorporates a retention teacher to mitigate the trade-off between concept erasure and preservation of generative capabilities. To further improve the coordination of multiple erasure objectives, we develop Adaptive Retention Control (ARC), which dynamically adjusts the sampling frequency and loss weight of each erasure teacher, together with the relative contribution of erasure and retention teachers throughout training. Extensive experiments on nudity, object, and artistic-style erasure demonstrate that FlowErase-OPD consistently improves the trade-off between erasure effectiveness, image quality, and semantic alignment, achieving state-of-the-art performance across diverse multi-concept erasure settings. Furthermore, the resulting models exhibit strong robustness against adversarial attacks. These results highlight the potential of on-policy distillation as a principled framework for safe and controllable generation in flow matching models.

Generating implementation-quality RTL with large language models (LLMs) remains difficult because direct generation must resolve microarchitecture while simultaneously producing and debugging low-level code. We present HINT, an executable hardware-intent intermediate representation layer between behavioral specifications or executable oracles and RTL. HINT makes RTL-relevant microarchitecture explicit, supports pre-RTL checking, and supplies explicit RTL-lowering obligations. We evaluate HINT using both a minimal single-agent flow and a full staged workflow. Across seven operator cases, the HINT-mediated route, with no post-synthesis QoR refinement, produces contract-compliant synthesizable RTL on 7/7 cases; Direct C2RTL and C2HLSC apply to five cases and succeed on 5/5 and 1/5, respectively. Under matched Design Compiler synthesis, HINT reduces area by 5.0%--26.2% relative to five manual RTL implementations and by 8.9%--86.1% relative to five accepted Direct C2RTL results. RealBench AES and SDC, together with a Vortex VPU synthesizing to 561.67k~$μ\mathrm{m}^2$, further demonstrate specification-driven, protocol-rich, memory-rich, and hierarchical designs. In the controlled operator study, the HINT-mediated route shows better observed convergence and avoids the severe implementation-quality degradation seen in several direct-generation results.

Aligning large language models (LLMs) with human values remains a major challenge, especially for trustworthy AI. While existing approaches such as RLHF, CAI, and their variants have achieved promising results, they often rely on a single-agent framework and a unified reward system. This limits their ability to capture ethical pluralism, adapt to diverse moral contexts, and reflect the dynamics of multi-agent moral reasoning. In this work, we propose a framework that utilizes multilayer combinatorial fusion for contextual value alignment (MCF-CVA). At the first layer of the framework, it instantiates multiple moral agents, each fine-tuned to represent a distinctive value. Their outputs are then expanded combinatorially using both score- and rank-combinations as well as average and weighted aggregations. These combined models are then reduced to the same number of initial moral agents. This expansion and reduction (EAR) process continues for multi-layers until a stopping criterion is reached. The MCF-CVA framework leverages cognitive diversity between agents to mitigate conflicts and redundancies across multiple agents, producing responses that better reflect contextual human values. The framework using the EAR algorithm is performed on the dual architecture of Euclidean score space and Kemeny rank space. Empirical evaluations demonstrated that the proposed framework outperforms single-agent baselines, multi-agent single-layer results, and previous aggregation approaches on standard metrics, showing that the MCF-CVA framework provides a robust and effective mechanism for advancing contextual value alignment in LLMs.

Agent Skills provide reusable capabilities to LLM agents. Agent Skill inconsistencies can expose undisclosed dangerous behavior or cause wrong Skill selection. Recent Agent Skill research has increasingly examined Agent Skill consistency detection. Existing methods evaluate behaviors or security-property graphs against predefined categories or declared scopes. More recently, PL-HCL uses an LLM-based model to learn consistency across metadata, instructions, and resources. However, declaration and implementation behavior can be mixed across text and code, and a concise declaration can correspond to multiple connected implementation steps. We present SkillConsist to address both challenges. An LLM separates declaration and implementation content into behavior records on the implementation and declaration sides, while static analysis supplements implementation records. These records form declaration and implementation behavior graphs, respectively. Starting from a behavior record on either side, bidirectional graph alignment searches the other graph for a candidate subgraph and expands it along behavior relations until it completely expresses the source-side behavior. Graph differencing identifies conflicts between aligned subgraphs and outputs the detection results. We construct a 633-Skill benchmark from ClawHub's 500 most-downloaded public Skills and 133 Skill-Inject packages. The benchmark contains 319 inconsistent and 314 consistent Skills and 442 localized inconsistency annotations. On this benchmark, SkillConsist achieves 86.85% precision, 89.03% recall, and 87.93% F1 for package-level detection, improving F1 over the best baseline by 20.43 percentage points. For localization, it achieves 67.60% precision, 58.14% recall, and 62.52% F1.

Long-term memory enables AI agents to maintain continuity across sessions, personalize behavior, and evolve through accumulated experience. Yet memory evolution is not simply a process of storing more information: new experiences may reinforce, revise, or interfere with existing memory states. Existing systems mainly emphasize memory construction and relevance-based retrieval, but several memories may remain simultaneously relevant while differing in state, temporal validity, or authority. We introduce Controlled Memory Interference (CMI), a controlled diagnostic and data-generation framework for studying how agent memory evolves under different memory relationships. Across controlled memory evolution, benign accumulation has limited effects, whereas relationship-specific interference sharply suppresses update plasticity with little stability gain, either by blocking target-memory exposure or by disrupting its downstream use. Lexical and Dense retrieval exhibit distinct interference pathways, while poisoning is more sensitive to update-authority cues than to recency alone. Beyond diagnosis, CMI provides targeted examples for interference-aware memory learning, improving the distinction between valid updates and interference-inducing memories while preserving performance on original memory tasks. These findings show that memory evolution is shaped not only by memory scale, but also by interactions among accumulated experiences. More broadly, memory interference emerges as an important factor for reliable continual agent memory systems.

Translating a natural-language question into a SPARQL query that can be executed against a large knowledge graph requires resolving lexical ambiguity, grounding surface terms in the target ontology, and producing graph patterns that are both syntactically valid and semantically faithful. We present an agentic text-to-SPARQL system that goes one step beyond static tool-using agents: a researcher agent that, after each round of inference on a validation set, proposes and tests changes to its own prompts, rules, and tool-orchestration code. We instantiate the loop on DBpedia, evolve nine successive versions of the agent driven by a low-cost reasoning model, and deploy the best-performing configuration with two stronger backbone models. The study yields three observations: (i) self-improvement converges quickly and then achieves 0.22 overall accuracy on the 2025 DBpedia validation set; (ii) the bottleneck is consistently in basic-graph-pattern predicate selection, not in SPARQL syntax or modifiers; and (iii) several benchmark items appear to penalise correct queries due to property ambiguity in DBpedia, suggesting that future Text-to-SPARQL benchmarks should be scored using a combination of machine translation and information retrieval metrics.

Large language models perform strongly on medical knowledge benchmarks, but reliable clinical deployment requires agents to conduct defensible investigations over heterogeneous, longitudinal records: determining what evidence is needed, retrieving and reconciling structured and free-text data, grounding conclusions in verifiable evidence, and deferring cases that cannot be resolved reliably. We introduce CliniCARE-Bench (Clinical Calibrated Audit of Medical Reasoning in EHR), a benchmark for retrospective clinical audit: 25 clinician-validated scenarios instantiated as 750 patient-specific cases over real-patient-derived MIMIC-IV data. Systems investigate each case through a governed, logged tool environment for record retrieval, computation, and policy access, and return one of four verdicts---Yes, No, Indeterminate: Lack of Data, or Indeterminate: Medically Ambiguous---the last two separating missing evidence from residual medical ambiguity. Beyond verdict accuracy, we score patient-evidence and policy grounding, process adherence, calibrated abstention, reliability, and efficiency against case-level reference verdicts produced by independent multi-model adjudication and calibrated against Clinical Board review. Every retrieval, computation, and report is replayable, so the investigation trace is inspectable and scorable. To our knowledge, CliniCARE-Bench is the first deployment-oriented clinical-agent benchmark to jointly evaluate real longitudinal EHR investigation, claim-level evidence grounding, governing-policy use, process adherence, and calibrated abstention within a common patient-level adjudication framework. Across 16 agentic systems, four-way accuracy spans 65.3-76.1%, but raw accuracy overstates investigation quality. Defect-free accuracy, which credits a verdict only when correct and free of prohibited shortcuts, is 4.8-14.8 points lower and reorders the leaderboard.

Hospitals are racing to embed AI, while coping with the surge in adaptation of the technology in other industries, into the triage management, documentation, scheduling, and revenue-cycle workflows, yet most deployments remain as fragmented pilots that stall at the edge of production, exposing patients and institutions to operational fragility, ungoverned risk, and mounting technical debt. At the same time, the global AI-in-healthcare market is projected to exceed nearly USD 1 trillion by 2034, according to the report of Fortune Business Insights, amplifying the financial consequences of architectural missteps and failed scaling strategies. This research proposes a compliance-first Agentic AI pattern catalogue and orchestration framework, purposely built for HIMS, moving beyond the single LLM chatbots and towards a governed ecosystem of autonomous and semi-autonomous agents. The framework extends by adding (i) a taxonomy of Agentic roles, (ii) a formal risk-stratification model that maps each pattern to risk tiers, human-in-the-loop checkpoints, and governance hooks, and (iii) a unified orchestration runtime capable of coordinating multi-agent workflows across EHR/HIMS landscapes such as Epic, Cerner, and MEDITECH. Technically the framework combines vLLM-based inference, optimized paging memory, confidential computing, and MCP based on-premise deployment, enforcing end-to-end encryption and policy-as-code controls aligned with HIPAA, GDPR, the EU AI Act, India's DPDP and DISHA Acts, ISO 27001, ISO 27002, ISO 14971 and IEC 62304. We exhibit how the proposed architecture is capable and efficient to reduce the documentation time, integration effort, and AI pilot attrition while constricting the governance and auditability, offering hospital leaders and governing authorities an urgently needed blueprint to convert AI investment into sustainable clinical, operational, and financial ROI

Graph Neural Networks (GNNs) are widely used across domains such as natural sciences, social network analysis, chip design, and recommendation systems. However, as graph sizes grow, storing and processing them entirely on a single-node CPU-GPU system becomes increasingly impractical. A promising approach is to distribute the graph across multiple remote memory nodes, though this introduces a major bottleneck: inter-node network congestion during training. To address this, we propose LGNNIC, a novel inter-node system architecture that leverages SmartNICs co-located with remote memory nodes-a configuration already available in modern systems-to reduce communication overhead in distributed GNN training. LGNNIC offloads key preprocessing tasks to SmartNICs, reducing the volume of data transferred to computational (training) nodes and alleviating network congestion. We introduce two complementary techniques executed on the SmartNICs during the preprocessing phase: Neighbor Sampling, which performs mini-batch sampling, and Quantization of the sampled batches. To evaluate LGNNIC under different communication infrastructures, we designed both an optimized low-overhead DMA-based synchronization mechanism and a high-overhead socket-based alternative used as a benchmark. We evaluate the core SmartNIC offloading mechanisms across standard GNN workloads and sampling hyperparameters using a proof-of-concept (PoC) system comprising one remote-memory node with an NVIDIA BlueField-2 SmartNIC and one compute node with an A100 GPU. Both Neighbor Sampling and Quantization on the remote node demonstrated substantial training speedups in most configurations. Neighbor Sampling achieved up to 62.4x and 17.5x speedups with Sockets and DOCA-DMA, respectively, primarily due to reduced data transaction time. Quantization provided additional speedups of up to 3.6x and 1.3x, respectively, by reducing data transfer.