Tech Business/VC TechCrunch

Gritt exits stealth with $34 million for robots to build solar plants—then, everything else

Robotics startup Gritt emerges from stealth with $34 million in funding to automate construction site tasks, specifically for solar plants.

Hardware/Chips The Verge

Who’s afraid of the big, bad GPU?

An exploration of the environmental and social costs of the AI GPU boom, including energy consumption, water usage, and e-waste.

AI/ML arXiv cs.AI

WuYu-EnvLE-Bench: A Benchmark for Evaluating Large Language Models in Environmental Law Enforcement

Introduction of WuYu-EnvLE-Bench, a benchmark for evaluating LLMs in environmental law enforcement cases.

Cybersecurity arXiv cs.AI

Dynamic Defense Profiling Enables Cognitive Jailbreak of Text-to-Image Models

MIND, a cognitive jailbreak framework for text-to-image models that models latent defense mechanisms to generate adversarial prompts.

AI/ML arXiv cs.AI

Financial Audit Assistance using Misinformation Detection and Explanation

A system using unsupervised techniques for misinformation detection and explanation in financial audit assistance.

AI/ML arXiv cs.AI

PGN: Design and Implementation of a Vision-Language Navigation System Based on Pangu Multimodal Foundation Model

PGN (Pangu Navigator), an offline vision-language navigation system based on the OpenPangu-7B multimodal foundation model.

Hardware/Chips arXiv cs.AI

A Hardware-oriented Approach for Efficient Bayesian Inference Computation and Deployment

A hardware-oriented approach to accelerate discrete Bayesian inference on embedded GPUs via memory layout restructuring and tensor clustering.

AI/ML arXiv cs.AI

Exploratory and Assimilating Reflection: Reflective Recall Cycle for Long-term Memory

The EAR framework introduces Exploratory and Assimilating Reflection to improve long-term memory retrieval for LLM-based autonomous agents.

AI/ML arXiv cs.AI

ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding

ST-Veto, a training-free method to improve diffusion multimodal LLM reasoning by vetoing temporally unstable and weakly grounded tokens.

AI/ML arXiv cs.AI

Mechanistic Attention Guidance for Agent Memory Refinement

Researchers propose AGMR, a framework that uses retrieval-head attention signals to guide targeted memory updates for AI agents, improving efficiency and performance over text-only methods.

AI/ML arXiv cs.AI

Verify, Repair, Repeat, or Stop? Robust Stopping for Noisy Verify-Repair Loops in LLM Agents

The VRR-Stop framework introduces a robust stopping mechanism for verify-repair loops in LLM agents using a noise model and belief filtering to prevent damaging correct plans.

AI/ML arXiv cs.AI

FlowBlock: Wavefront-Parallel Decoding for Self-Correcting Diffusion Language Models

FlowBlock introduces wavefront-parallel decoding for self-correcting diffusion language models, significantly increasing tokens per second and reducing latency without requiring retraining.

AI/ML arXiv cs.AI

OrientSAM: Mitigating Camera-Centric Shortcut in Multimodal Spatial Reasoning via Orientation-Aware Spatial Alignment

OrientSAM is a framework designed to improve multimodal spatial reasoning in LLMs by injecting explicit orientation information via Fourier-based angle encoding.

AI/ML arXiv cs.AI

Artificial Intelligence for Understanding and Managing Transportation Behavior in Sustainable Smart Cities

A study explores the application of AI for managing urban transportation behavior in smart cities, emphasizing a behavior-centered perspective on mobility data.

AI/ML arXiv cs.AI

ProEvent: An Event-centric Benchmark for Proactive Agents

ProEvent is a new event-centric benchmark to evaluate the ability of proactive AI agents to maintain user timetables from chat interactions, revealing current LLM limitations.

AI/ML arXiv cs.AI

LaT: LLM-as-Trainer for Multi-Task Vehicle Routing Solvers

The LLM-as-Trainer (LaT) paradigm uses a pretrained LLM as an external trainer to provide stage-wise guidance for multi-task vehicle routing solvers.

AI/ML arXiv cs.AI

Learning to Detect Cross-Modal Negation: An Analysis of Latent Representations and an Attention-Based Solution

Researchers analyze cross-modal negation detection in vision-language models and propose an attention-based architecture to better model inter-modal dependencies.

AI/ML arXiv cs.AI

SR-Agent: An Experience-Driven Agentic Framework for Post-Ranking Strategies Refinement in E-Commerce Recommendation

SR-Agent is an experience-driven framework for automating the refinement of post-ranking strategies in industrial e-commerce recommender systems.

AI/ML arXiv cs.AI

Semantically Similar, Logically Distinct: Diagnosing the Semantic-Answerability Gap in Table RAG

The TCR-Bench benchmark diagnoses the 'Semantic-Answerability Gap' in Table RAG, showing that semantic relevance does not guarantee that a table can actually answer a query.

AI/ML Hacker News

Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge

Alibaba introduces Qwen-Image-3.0, a multimodal model focusing on high-fidelity image understanding and detailed knowledge extraction.