Paper page - XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural Sensitivity
… Evaluating 10 frontier and 27 local LLMs reveals two key findings. …
Search
Paper page - RemoteZero: Geospatial Reasoning with Zero Human Annotations
… The following papers were recommended by the Semantic Scholar API RemoteAgent: Bridging Vague Human Intents and Earth Observation with RL-based Agentic MLLMs 2026 GeoSolver: Scaling Test-Time Reasoning in Remote Sensing with Fine-Grained Process Supervision 2026 Think and Answer ME: Benchmarking an… …
Paper page - LoMo: Local Modality Substitution for Deeper Vision-Language Fusion
Papers arxiv:2605.30265 LoMo: Local Modality Substitution for Deeper Vision-Language Fusion Published on May 28 Submitted by Zhixiong Zhang SII on May 29 Fudan University Authors: , Zhixiong Zhang , , Yibin Wang , Abstract Vision-language models suffer from modality sensitivity due to training data… …
Paper page - From Prompt Injection to Persistent Control: Defending Agentic Harness Against Trojan Backdoors
… In local agentic harnesses, an LLM can read and write files, call tools, and reuse workspace state across sessions. …
Paper page - RADIO-ViPE: Online Tightly Coupled Multi-Modal Fusion for Open-Vocabulary Semantic SLAM in Dynamic Environments
… AI-generated summary We present RADIO-ViPE Reduce All Domains Into One -- Video Pose Engine , an online semantic SLAM system that enables geometry-aware open-vocabulary grounding , associating arbitrary natural language queries with localized 3D regions and objects in dynamic environments . …
Paper page - LaRA: Layer-wise Representation Analysis for Detecting Data Contamination in RL Post-Training
… LaRA introduces three complementary metrics, measuring perturbation sensitivity , directional collapse , and local representation rigidity under controlled perturbations. …
Paper page - Count Anything
… This formulation unifies category-conditioned counting with interpretable spatial localization. …
Paper page - SOCO: Benchmarking Semantic Object Correspondence in Vision Foundation Models
… Comprehensive experiments reveal that i vision foundation backbones encode strong semantic structure but transfer correspondences poorly across related categories and only partially capture object-part position, ii LVLMs are stronger at text-prompted part localization than at visual-reference cross… …
Paper page - When No Benchmark Exists: Validating Comparative LLM Safety Scoring Without Ground-Truth Labels
… We instantiate the chain in SimpleAudit, a local-first scoring instrument, and validate it on a Norwegian safety pack. …
Paper page - Learning to Act and Cooperate for Distributed Black-Box Consensus Optimization
… Potential applications include cooperative sensing, resource allocation, and distributed control, which may contribute to more efficient and resilient large-scale systems. …