PlexRL: Cluster-Level Orchestration of Serviceized LLM Execution for RLVR

Reinforcement learning with verifiable rewards (RLVR) has recently unlocked strong reasoning capabilities in large language models (LLMs), triggering rapid exploration of new algorithms and data. However, RLVR training is notoriously inefficient: long-tailed rollouts, tool-induced stalls, and asymmetric resource requirements between rollout and training introduce substantial idle time that cannot be eliminated by job-local optimizations such as synchronous pipelining, asynchronous rollout, or colocated execution. We argue that this inefficiency is structural. While idle gaps are unavoidable within individual RLVR jobs, they are largely anti-correlated across jobs and therefore exploitable at the cluster level. Leveraging this observation, we present PlexRL, a cluster-level runtime for multiplexing unified LLM services across RLVR jobs. By centrally managing model placement, state transitions, and function-level scheduling under strict affinity constraints, PlexRL time-slices LLM execution across jobs to fill otherwise idle periods without expensive model migration. Our implementation and evaluations demonstrate that PlexRL significantly improves effective cluster capacity and reduces user GPU hour cost by maximum 37.58% while preserving algorithmic flexibility and introducing minimal per-job overhead.

When Cases Get Rare: A Retrieval Benchmark for Off-Guideline Clinical Question Answering

Across medical specialties, clinical practice is anchored in evidence-based guidelines that codify best studied diagnostic and treatment pathways. These pathways routinely fall short for the long tail of real-world care not covered by guidelines. Most medical large language models (LLMs), however, are trained to encode common, guideline-focused medical knowledge in their parameters. Current evaluations test models primarily on recalling and reasoning with this memorized content, often in multiple-choice settings. Given the fundamental importance of evidence-based reasoning in medicine, it is neither feasible nor reliable to depend on memorization in practice. To address this gap, we introduce OGCaReBench, a free-form retrieval-focused benchmark aimed at evaluating LLMs at answering clinical questions that require going beyond typical guidelines. Extracted from published medical case reports and validated by medical experts, OGCaReBench contains long-form clinical questions requiring free-text answers, providing a systematic framework for assessing open-ended medical reasoning in rare, case-based scenarios. Our experiments reveal that even the best-performing baseline (GPT-5.2) correctly answers only 56% of our benchmark with specialized models only reaching 42%. Augmenting models with retrieved medical articles improves this performance to up to 82% (using GPT-5.2) highlighting the importance of evidence-grounding for real-world medical reasoning tasks. This work thus establishes a foundation for benchmarking and advancing both general-purpose and medical LLMs to produce reliable answers in challenging clinical contexts.

ChannelAgent-Empowered Electromagnetic Space World Model: A Case Study on Agent-Driven Channel Generation for 6G AI-Native Air Interface

As sixth-generation (6G) wireless networks evolve toward increasingly heterogeneous scenarios, tasks, and service requirements, conventional artificial intelligence (AI) models remain limited in task-aware decision-making and autonomous adaptation. To address this issue, this paper first proposes a ChannelAgent-empowered electromagnetic space world model, in which wireless intelligence is organized into a closed-loop process consisting of multi-modal sensing, ChannelAgent as the intelligent core, and execution with feedback update. As a case study, agent-driven channel generation is instantiated through path loss prediction. Specifically, a task-oriented intelligent feature selection mechanism is designed by integrating reinforcement-learning-inspired policy adaptation with evolutionary search, enabling the agent to iteratively derive compact and task-suitable feature subsets according to the current scenario and performance feedback. Simulation results demonstrate superior performance in both single-scenario and multi-scenario tasks, highlighting the potential of the proposed model for autonomous, adaptive, task-oriented, and closed-loop wireless intelligence.

Deco: Extending Personal Physical Objects into Pervasive AI Companion through a Dual-Embodiment Framework

Individuals frequently form deep attachments to physical objects (e.g., plush toys) that usually cannot sense or respond to their emotions. While AI companions offer responsiveness and personalization, they exist independently of these physical objects and lack an ongoing connection to them. To bridge this gap, we conducted a formative study (N=9) to explore how digital agents could inherit and extend the emotional bond, deriving four design principles (Faithful Identity, Calibrated Agency, Ambient Presence, and Reciprocal Memory). We then present the Dual-Embodiment Companion Framework, instantiated as Deco, a mobile system integrating multimodal Large Language Models (LLMs) and Augmented Reality to create synchronized digital embodiments of users' physical companions. A within-subjects study (N=25) showed Deco significantly outperformed a personalized LLM-empowered digital companion baseline on perceived companionship, emotional bond, and design-principle scales (all p<0.01). A seven-day field deployment (N=17) showed sustained engagement, subjective well-being improvement (p=.040), and three key relational patterns: digital activities retroactively vitalized physical objects, bond deepening was driven by emotional engagement depth rather than interaction frequency, and users sustained bonds while actively navigating digital companions' AI nature. This work highlights a promising alternative for designing digital companions: moving from creating new relationships to dual embodiment, where digital agents seamlessly extend the emotional history of physical objects.

IPRU: Input-Perturbation-based Radio Frequency Fingerprinting Unlearning for LAWNs

Radio Frequency Fingerprinting (RFF) is a key technology for identity authentication in wireless networks. However, due to the rapid dynamics of Autonomous Aerial Vehicles (AAVs) in low-altitude wireless networks, RFF models require parameter updates to maintain authentication performance, posing a major challenge to existing schemes. Conventional retraining approaches for handling departed or compromised AAVs are computationally prohibitive and risk retaining polluted features, which compromises both authentication security and user privacy. To address these limitations, we propose an Input-Perturbation-based RFF Unlearning (IPRU) scheme. By optimizing a universal Fingerprint Forget Vector (FFV) as a lightweight input perturbation, IPRU successfully erases the fingerprints of target AAVs without modifying the RFF model parameters, achieving an effective balance between efficient unlearning and preserved authentication performance. A combinatorial optimization strategy further enables multi-AAV forgetting on demand. The simulation results demonstrate that IPRU achieves 1.41% unlearning accuracy, 99.41% remaining accuracy, and 100% resistance to membership inference attack, while running 5.79X faster than retraining and 2.1X faster than the baseline scheme.

PBSBench: A Multi-Level Vision-Language Framework and Benchmark for Hematopathology Whole Slide Image Interpretation

Peripheral Blood Smear (PBS) is a critical microscopic examination in hematopathology that yields whole-slide imaging (WSI). Unlike solid tissue pathology, PBS interpretation focuses on individual cell morphologies rather than tissue architecture, making it distinct in both visual characteristics and diagnostic reasoning. However, current multimodal large language models (MLLMs) for pathology are primarily developed on solid-tissue WSIs and struggle to generalize to PBS. To bridge this gap, we construct PBSInstr, the first vision-language dataset for PBS interpretation, comprising 353 PBS WSIs paired with microscopic impression paragraphs and 29k cell-level image crops annotated with cell type labels and morphological descriptions. To facilitate instruction tuning, PBSInstr further includes 27k question-answer (QA) pairs for cell crops and 1,286 QA pairs for PBS slides. Building upon PBSInstr, we develop PBS-VL, a hematopathology-tailored vision-language model for multi-level PBS interpretation at both cell and slide levels. To comprehensively evaluate PBS understanding, we construct PBSBench, a visual question answering (VQA) benchmark featuring four question categories and six PBS interpretation tasks. Experiments show that PBS-VL outperforms existing general-purpose and pathology MLLMs, underscoring the value of PBS-specific data. We release our code, datasets, and model weights to facilitate future research. Our proposed framework lays the foundation for developing practical AI assistants supporting decision-making in hematopathology.

A Synonymous Variational Perspective on the Rate-Distortion-Perception Tradeoff

The fundamental limit of natural signal compression has traditionally been characterized by classical rate-distortion (RD) theory through the tradeoff between coding rate and reconstruction distortion, while the rate-distortion-perception (RDP) framework introduces a divergence-based measure of perceptual quality as a modeling principle rather than a theoretically-derived principle, leaving its theoretical origin unclear. In this paper, motivated by a synonymity-based semantic information perspective, we reformulate perceptual reconstruction as recovering any admissible sample within an ideal synonymous set (synset) associated with the source, rather than the source sample itself, and correspondingly establish a synonymous source coding architecture. On this basis, we develop a synonymous variational inference (SVI) analysis framework with a synonymous variational lower bound (SVLBO) for tractable analysis of synset-oriented compression. Within this framework, we establish a synonymity-perception consistency principle, showing that optimal identification of semantic information is theoretically consistent with perceptual optimization. Based on its derivation result, we prove a synonymous RDP tradeoff for the proposed synonymous source coding. These analytical results show that the distributional divergence term arises naturally from the synset-based reconstruction objective, clarify its compatibility with existing RDP formulations and classical RD theory, and suggest the potential advantages of synonymous source coding.

CoRe-ECG: Advancing Self-Supervised Representation Learning for 12-Lead ECG via Contrastive and Reconstructive Synergy

Accurate interpretation of electrocardiogram (ECG) remains challenging due to the scarcity of labeled data and the high cost of expert annotation. Self-supervised learning (SSL) offers a promising solution by enabling models to learn expressive representations from unlabeled signals. Existing ECG SSL methods typically rely on either contrastive learning or reconstructive learning. However, each approach in isolation provides limited supervisory signals and suffers from additional limitations, including non-physiological distortions introduced by naive augmentations and trivial correlations across multiple leads that models may exploit as shortcuts. In this work, we propose CoRe-ECG, a unified contrastive and reconstructive pretraining paradigm that establishes a synergistic interaction between global semantic modeling and local structural learning. CoRe-ECG aligns global representations during reconstruction, enabling instance-level discriminative signals to guide local waveform recovery. To further enhance pretraining, we introduce Frequency Dynamic Augmentation (FDA) to adaptively perturb ECG signals based on their frequency-domain importance, and Spatio-Temporal Dual Masking (STDM) to break linear dependencies across leads, increasing the difficulty of reconstructive tasks. Our method achieves state-of-the-art performance across multiple downstream ECG datasets. Ablation studies further demonstrate the necessity and complementarity of each component. This approach provides a robust and physiologically meaningful representation learning framework for ECG analysis.

MathGen: Revealing the Illusion of Mathematical Competence through Text-to-Image Generation

Modern generative models have demonstrated the ability to solve challenging mathematical problems. In many real-world settings, however, mathematical solutions must be expressed visually through diagrams, plots, geometric constructions, and structured symbolic layouts, where correctness depends on precise visual composition. This naturally raises the question of whether generative models can still do so when the answer must be rendered visually rather than written in text? To study this problem, we introduce MathGen, a rigorous benchmark of 900 problems spanning seven core domains, each paired with an executable verifier under a Script-as-a-Judge protocol for deterministic and objective evaluation. Experiments on representative open-source and proprietary text-to-image models show that mathematical fidelity remains a major bottleneck: even the best closed-source model reaches only 42.0% overall accuracy, while open-source models achieve just ~ 1-11%, often near 0% on structured tasks. Overall, current T2I models remain far from competent at even elementary mathematical visual generation.

Joint Sensing and Covert Communications in RIS-NOMA Systems

A reconfigurable intelligent surface (RIS)-assisted non-orthogonal multiple access (NOMA) system is investigated, where the transmitter (Alice) is a dual functional radar communication (DFRC) base station (BS) that aims to sense the location of a potential warden (Willie), while simultaneously transmitting public and covert signals to the legitimate users, Carol and Bob, respectively. Both cases of known and unknown Willie locations are considered. For the known-location case, assuming perfect channel state information (CSI) at Willie, a covert rate maximization is formulated with the joint optimization of active and passive beamforming, which is solved using successive convex approximation (SCA), penalty method, and semidefinite relaxation (SDR). For the unknown-location case, we propose to estimate Willie's location via radar sensing and develop a sensing-based imperfect CSI model. In particular, the CSI error uncertainty is bounded by the sensing accuracy, which is characterized by the Cramer-Rao bound (CRB). Subsequently, a robust communication rate maximization problem is formulated under the constraints on quality-of-service (QoS) of Carol, sensing accuracy, and covertness level. The Schur complement and S-procedure are employed to handle the non-convex constraints. Numerical results compare the system performance under the two cases, and demonstrate the significant covert performance superiority of the sensing-based imperfect CSI model and NOMA over the general norm-bounded imperfect CSI model and the orthogonal multiple access scheme. Furthermore, the dual yet contradictory effects of sensing on covert communications are revealed. It is also found that Alice primarily utilizes Carol's signal for sensing, while allocating almost all of Bob's signal for communication.