arXiv:2604.14156v1 Announce Type: new Abstract: Large language models deliver strong generative performance but at the cost of massive parameter counts, memory use, and decoding latency. Prior work has shown that pruning and structured sparsity can preserve accuracy under substantial compression, while prompt-compression methods reduce latency by removing redundant input tokens. However, these two directions remain largely separate. Most model-compression methods are static and optimized offline, and they do not exploit the fact that different prompts and decoding steps activate different latent computational pathways. Prompt-compression methods reduce sequence length, but they do not adapt the executed model subnetwork. We propose a unified compressed-sensing-guided framework for dynamic LLM execution. Random measurement operators probe latent model usage, sparse recovery estimates task-conditioned and token-adaptive support sets, and the recovered supports are compiled into hardware-efficient sparse execution paths over blocks, attention heads, channels, and feed-forward substructures. The framework introduces five key contributions: task-conditioned measurements, so different prompts induce different sparse supports; token-adaptive recovery, so active substructures are re-estimated during decoding; formal sample-complexity bounds under restricted isometry or mutual incoherence assumptions; compile-to-hardware constraints that restrict recovery to GPU-efficient structures; and a joint objective that unifies prompt compression with model reduction. Together, these components recast LLM inference as a measurement-and-recovery problem with explicit approximation guarantees and deployment-oriented speedup constraints.

arXiv:2604.14329v1 Announce Type: new Abstract: Non-violent street robberies (snatch-and-run) are difficult to detect automatically because they are brief, subtle, and often indistinguishable from benign human interactions in unconstrained surveillance footage. This paper presents a hybrid, pose-driven approach for detecting snatch-and-run events that combines real-time perception with an interpretable classification stage suitable for edge deployment. The system uses a YOLO-based pose estimator to extract body keypoints for each tracked person and computes kinematic and interaction features describing hand speed, arm extension, proximity, and relative motion between an aggressor-victim pair. A Random Forest classifier is trained on these descriptors, and a temporal hysteresis filter is applied to stabilize frame-level predictions and reduce spurious alarms. We evaluate the method on a staged dataset and on a disjoint test set collected from internet videos, demonstrating promising generalization across different scenes and camera viewpoints. Finally, we implement the complete pipeline on an NVIDIA Jetson Nano and report real-time performance, supporting the feasibility of proactive, on-device robbery detection.

arXiv:2604.14161v1 Announce Type: new Abstract: Reliable evaluation is essential in machine learning research, yet methodological flaws-particularly data leakage-continue to undermine the validity of reported results. In this work, we investigate whether large language models (LLMs) can act as independent analytical agents capable of identifying such issues in published studies. As a case study, we analyze a gesture-recognition paper reporting near-perfect accuracy on a small, human-centered dataset. We first show that the evaluation protocol is consistent with subject-level data leakage due to non-independent training and test splits. We then assess whether this flaw can be detected independently by six state-of-the-art LLMs, each analyzing the original paper without prior context using an identical prompt. All models consistently identify the evaluation as flawed and attribute the reported performance to non-independent data partitioning, supported by indicators such as overlapping learning curves, minimal generalization gap, and near-perfect classification results. These findings suggest that LLMs can detect common methodological issues based solely on published artifacts. While not definitive, their consistent agreement highlights their potential as complementary tools for improving reproducibility and supporting scientific auditing.

arXiv:2604.14163v1 Announce Type: new Abstract: Maritime distress communications transmitted over very high frequency (VHF) radio are safety-critical voice messages used to report emergencies at sea. Under the Global Maritime Distress and Safety System (GMDSS), such messages follow standardized procedures and are expected to convey essential details, including vessel identity, position, nature of the distress, and required assistance. In practice, however, automatic analysis remains difficult because distress messages are often brief, noisy, and produced under stress, may deviate from the prescribed format, and are further degraded by automatic speech recognition (ASR) errors caused by channel noise and speaker stress. This paper presents SeaAlert, an LLM-based framework for robust analysis of maritime distress communications. To address the scarcity of labeled real-world data, we develop a synthetic data generation pipeline in which an LLM produces realistic and diverse maritime messages, including challenging variants in which standard distress codewords are omitted or replaced with less explicit expressions. The generated utterances are synthesized into speech, degraded with simulated VHF noise, and transcribed by an ASR system to obtain realistic noisy transcripts.

arXiv:2604.14268v1 Announce Type: new Abstract: We introduce HY-World 2.0, a multi-modal world model framework that advances our prior project HY-World 1.0. HY-World 2.0 accommodates diverse input modalities, including text prompts, single-view images, multi-view images, and videos, and produces 3D world representations. With text or single-view image inputs, the model performs world generation, synthesizing high-fidelity, navigable 3D Gaussian Splatting (3DGS) scenes. This is achieved through a four-stage method: a) Panorama Generation with HY-Pano 2.0, b) Trajectory Planning with WorldNav, c) World Expansion with WorldStereo 2.0, and d) World Composition with WorldMirror 2.0. Specifically, we introduce key innovations to enhance panorama fidelity, enable 3D scene understanding and planning, and upgrade WorldStereo, our keyframe-based view generation model with consistent memory. We also upgrade WorldMirror, a feed-forward model for universal 3D prediction, by refining model architecture and learning strategy, enabling world reconstruction from multi-view images or videos. Also, we introduce WorldLens, a high-performance 3DGS rendering platform featuring a flexible engine-agnostic architecture, automatic IBL lighting, efficient collision detection, and training-rendering co-design, enabling interactive exploration of 3D worlds with character support. Extensive experiments demonstrate that HY-World 2.0 achieves state-of-the-art performance on several benchmarks among open-source approaches, delivering results comparable to the closed-source model Marble. We release all model weights, code, and technical details to facilitate reproducibility and support further research on 3D world models.

arXiv:2604.14164v1 Announce Type: new Abstract: A widely adopted strategy for model enhancement is to use synthetic data generated by a stronger model for supervised fine-tuning (SFT). However, for emerging reasoning models like Qwen3-8B, this approach often fails to improve reasoning capabilities and can even lead to a substantial drop in performance. In this work, we identify substantial stylistic divergence between teacher generated data and the distribution of student as a major factor impacting SFT. To bridge this gap, we propose a Teacher-Student Cooperation Data Synthesis framework (TESSY), which interleaves teacher and student models to alternately generate style and non-style tokens. Consequently, TESSY produces synthetic sequences that inherit the advanced reasoning capabilities of the teacher while maintaining stylistic consistency with the distribution of the student. In experiments on code generation using GPT-OSS-120B as the teacher, fine-tuning Qwen3-8B on teacher-generated data leads to performance drops of 3.25% on LiveCodeBench-Pro and 10.02% on OJBench, whereas TESSY achieves improvements of 11.25% and 6.68%.

arXiv:2604.14193v1 Announce Type: new Abstract: Human 3D vision involves two distinct stages: an Experience Module, where stereo depth is extracted relative to fixation, and an Inference Module, where this experience is interpreted to estimate 3D scene properties. Paradoxically, although our experience of stereo vision does not provide us with distance information, it does affect our inferences about visual scale. We propose the Inference Module exploits a natural scene statistic: near scenes produce vivid disparity gradients, while far scenes appear comparatively flat. QualiaNet implements this two-stage architecture computationally: disparity maps simulating human stereo experience are passed to a CNN trained to estimate distance. The network can recover distance from disparity gradients alone, validating this approach.

arXiv:2604.14165v1 Announce Type: new Abstract: We present EviSearch, a multi-agent extraction system that automates the creation of ontology-aligned clinical evidence tables directly from native trial PDFs while guaranteeing per-cell provenance for audit and human verification. EviSearch pairs a PDF-query agent (which preserves rendered layout and figures) with a retrieval-guided search agent and a reconciliation module that forces page-level verification when agents disagree. The pipeline is designed for high-precision extraction across multimodal evidence sources (text, tables, figures) and for generating reviewer-actionable provenance that clinicians can inspect and correct. On a clinician-curated benchmark of oncology trial papers, EviSearch substantially improves extraction accuracy relative to strong parsed-text baselines while providing comprehensive attribution coverage. By logging reconciler decisions and reviewer edits, the system produces structured preference and supervision signals that bootstrap iterative model improvement. EviSearch is intended to accelerate living systematic review workflows, reduce manual curation burden, and provide a safe, auditable path for integrating LLM-based extraction into evidence synthesis pipelines.

arXiv:2604.14314v1 Announce Type: new Abstract: This manuscript introduces DharmaOCR Full and Lite, a pair of specialized small language models (SSLMs) for structured OCR that jointly optimize transcription quality, generation stability, and inference cost. It also presents DharmaOCR-Benchmark, a benchmark that covers printed, handwritten, and legal/administrative documents, and proposes a unified evaluation protocol that measures fidelity and structure while explicitly tracking text degeneration as a first-class benchmark metric (alongside unit cost). Beyond reporting degeneration rates, the manuscript empirically shows degeneration is not merely a quality failure, since it materially worsens production performance by increasing response time, reducing throughput, and inflating computational cost due to abnormally long generations. To the best of the author's knowledge, as a methodological contribution, this is the first application of Direct Preference Optimization (DPO) for OCR, explicitly using degenerate generations as rejected examples to penalize looping behavior. Combined with Supervised Fine-Tuning (SFT) for enforcing a strict JSON schema (header, margin, footer, and text), DPO consistently reduces degeneration rate across model families (up to 87.6% relative) while preserving or improving extraction quality. The resulting models, namely, DharmaOCR Full (7B) and DharmaOCR Lite (3B), set a new state-of-the-art on DharmaOCR-Benchmark, outperforming each open-source and commercial baseline model evaluated regarding extraction quality, reaching 0.925 and 0.911 scores with 0.40% and 0.20% degeneration rates. AWQ quantization reduced up to 22% per-page cost with negligible quality loss, enabling a strong quality-cost trade-off in comparison to proprietary OCR APIs and open-source alternatives.

arXiv:2604.14433v1 Announce Type: new Abstract: Zero-ablation -- replacing token activations with zero vectors -- is widely used to probe token function in vision transformers. Register zeroing in DINOv2+registers and DINOv3 produces large drops (up to $-36.6$\,pp classification, $-30.9$\,pp segmentation), suggesting registers are functionally indispensable. However, three replacement controls -- mean-substitution, noise-substitution, and cross-image register-shuffling -- preserve performance across classification, correspondence, and segmentation, remaining within ${\sim}1$\,pp of the unmodified baseline. Per-patch cosine similarity shows these replacements genuinely perturb internal representations, while zeroing causes disproportionately large perturbations, consistent with why it alone degrades tasks. We conclude that zero-ablation overstates dependence on exact register content. In the frozen-feature evaluations we test, performance depends on plausible register-like activations rather than on exact image-specific values. Registers nevertheless buffer dense features from \texttt{[CLS]} dependence and are associated with compressed patch geometry. These findings, including the replacement-control results, replicate at ViT-B scale.

arXiv:2604.14373v1 Announce Type: new Abstract: Rural environmental risks are shaped by place-based conditions (e.g., housing quality, road access, land-surface patterns), yet standard vulnerability indices are coarse and provide limited insight into risk contexts. We propose SatBLIP, a satellite-specific vision-language framework for rural context understanding and feature identification that predicts county-level Social Vulnerability Index (SVI). SatBLIP addresses limitations of prior remote sensing pipelines-handcrafted features, manual virtual audits, and natural-image-trained VLMs-by coupling contrastive image-text alignment with bootstrapped captioning tailored to satellite semantics. We use GPT-4o to generate structured descriptions of satellite tiles (roof type/condition, house size, yard attributes, greenery, and road context), then fine-tune a satellite-adapted BLIP model to generate captions for unseen images. Captions are encoded with CLIP and fused with LLM-derived embeddings via attention for SVI estimation under spatial aggregation. Using SHAP, we identify salient attributes (e.g., roof form/condition, street width, vegetation, cars/open space) that consistently drive robust predictions, enabling interpretable mapping of rural risk environments.

arXiv:2604.14388v1 Announce Type: new Abstract: Humans routinely infer taste, smell, texture, and even sound from food images a phenomenon well studied in cognitive science. However, prior vision language research on food has focused primarily on recognition tasks such as meal identification, ingredient detection, and nutrition estimation. Image-based prediction of multisensory experience remains largely unexplored. We introduce FoodSense, a human-annotated dataset for cross-sensory inference containing 66,842 participant-image pairs across 2,987 unique food images. Each pair includes numeric ratings (1-5) and free-text descriptors for four sensory dimensions: taste, smell, texture, and sound. To enable models to both predict and explain sensory expectations, we expand short human annotations into image-grounded reasoning traces. A large language model generates visual justifications conditioned on the image, ratings, and descriptors. Using these annotations, we train FoodSense-VL, a vision language benchmark model to produce both multisensory ratings and grounded explanations directly from food images. This work connects cognitive science findings on cross-sensory perception with modern instruction tuning for multimodal models and shows that many popular evaluation metrics are insufficient for visually sensory inference.

Europe's first Mars rover mission is now on its fourth rocket: SpaceX's Falcon Heavy

Kyle Orland / Ars Technica: Mozilla launches Thunderbolt, an open-source AI client for users and businesses who want to run their own self-hosted AI infrastructure, available on GitHub  —  Mozilla is the latest legacy tech brand to make a play for the enterprise AI market.  But the company behind Firefox and Thunderbird …

The fundraise is the first major capital raise under Sequoia's new leadership, with Alfred Lin and Pat Grady now serving as co-stewards of the 54-year-old firm.

Nilay Patel / The Verge: Q&A: Ronan Farrow on his Sam Altman profile in The New Yorker, how Altman changed in 18 months of reporting, OpenAI's board drama and enterprise shift, and more  —  Today on Decoder, I'm talking with Ronan Farrow, one of the biggest stars of investigative reporting working today.

The Information: Sources: Cerebras plans to make its IPO public as soon as Friday, aiming to raise $3B+ at a $35B+ valuation, a 60% premium to its $22B February valuation  —  Chip startup Cerebras is planning to make its IPO paperwork public as soon as Friday, The Information reported Thursday.

The Information: Sources: OpenAI agrees to pay Cerebras $20B+ to use its server chips, double the amount previously associated with the deal, and may receive equity in Cerebras  —  OpenAI recently struck an unusual deal to lessen its dependence on Nvidia's AI chips and potentially lower its computing expenses in the coming years.

Jessica Hill / Associated Press: Roblox agrees to pay $12.5M and implement age verification for all users as part of a settlement with Nevada over claims it failed to protect young users  —  Roblox, a gaming platform popular with kids, will implement increased protections for young users and pay more than $12 million …

Bloomberg: Sources: SpaceX told employees that it has moved up the scheduled date for when shares awarded to employees become eligible for sale from May to April  —  SpaceX has moved up a scheduled vesting date for shares awarded to employees to as soon as next week, according to people familiar with the matter …