Actualités
GuidedBench: Measuring and Mitigating the Evaluation Discrepancies of In-the-wild LLM Jailbreak Methods
arXiv:2502.16903v2 Announce Type: replace Abstract: Despite the growing interest in jailbreak methods as an effective...
Graph-R1: An Agentic GraphRAG Framework for Structured, Multi-Turn Reasoning with Reinforcement Learning
Introduction Large Language Models (LLMs) have set new benchmarks in natural language processing, but their...
Graph-Based Spectral Decomposition for Parameter Coordination in Language Model Fine-Tuning
arXiv:2504.19583v2 Announce Type: replace-cross Abstract: This paper proposes a parameter collaborative optimization algorithm for large...
GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs
arXiv:2507.18043v1 Announce Type: new Abstract: Inference-time steering methods offer a lightweight alternative to fine-tuning large...
GPZ: A Next-Generation GPU-Accelerated Lossy Compressor for Large-Scale Particle Data
Particle-based simulations and point-cloud applications are driving a massive expansion in the size and complexity...
GPT-4o Understands Text, But Does It See Clearly? A Benchmarking Study of MFMs on Vision Tasks
Multimodal foundation models (MFMs) like GPT-4o, Gemini, and Claude have shown rapid progress recently, especially...
GPT and Prejudice: A Sparse Approach to Understanding Learned Representations in Large Language Models
arXiv:2510.01252v2 Announce Type: replace Abstract: As large language models (LLMs) are increasingly trained on massive...
Google’s Jules aims to out-code Codex in battle for the AI developer stack
Google released Jules, its coding agent, into beta as autonomous coding agents are quickly gaining...
Google’s AlphaEvolve: The AI agent that reclaimed 0.7% of Google’s compute – and how to copy it
Google’s AlphaEvolve is the epitome of a best-practice AI agent orchestration. It offers a lesson...
Google’s ‘world-model’ bet: building the AI operating layer before Microsoft captures the UI
Google doubles down on its ‘world-model’ vision, racing to build an AI operating layer to...
Google Researchers Introduce LightLab: A Diffusion-Based AI Method for Physically Plausible, Fine-Grained Light Control in Single Images
Manipulating lighting conditions in images post-capture is challenging. Traditional approaches rely on 3D graphics methods...
Google releases Olympiad medal-winning Gemini 2.5 ‘Deep Think’ AI publicly — but there’s a catch…
The Gemini 2.5 Deep Think released to users is not that same competition model, rather...
