AI
When Google researchers first introduced SigLIP in 2023, it reimagined vision-language training by replacing CLIP’s contrastive learning with a simpler binary classification approach. Now, the team is back with SigLIP 2—a family of open models that pushes multimodal AI forward through architectural tweaks, smarter training strategies, and a
A new open-source multimodal AI called Ovis2 is turning heads by matching—and sometimes surpassing—the capabilities of models ten times its size. Developed by AIDC-AI, this 34-billion parameter system builds on last year's Ovis1.6 architecture with structural improvements that better align visual and textual understanding, while
Alibaba Cloud has unveiled a major upgrade to its vision-language AI system that could reshape how enterprises handle everything from logistics paperwork to security footage analysis. Qwen2.5-VL, the newest iteration of the company’s multimodal model series, builds on its predecessor’s ability to interpret images and text with
New multi-agent system built on Gemini 2.0 accelerates discovery, but human oversight remains critical. In a bid to accelerate scientific discovery, Google researchers have unveiled an AI "co-scientist" that autonomously generates biomedical hypotheses, debates their merits, and iteratively refines them—a process culminating in lab-validated findings, including
As large language models push into territory once reserved for human cognition—analyzing entire code repositories, summarizing novels, or maintaining coherent conversations spanning hours—the computational demands of traditional attention mechanisms have become unsustainable. Chinese AI lab DeepSeek’s newly proposed Native Sparse Attention (NSA) tackles this challenge through a
Elon Musk’s xAI has unveiled Grok 3, a new large language model positioned as a competitor to OpenAI’s GPT-4o, Google’s Gemini, and China’s DeepSeek. The company claims the model achieves “superhuman reasoning” through a combination of architectural upgrades and synthetic training data, though independent researchers urge
This article was originally written in English. As an experiment and for your own assessment, we have included translations provided by Mistral Saba هذا المقال كتب أصلاً باللغة الإنجليزية. كتجربة ولتقييمك الخاص، قمنا بتضمين الترجمات التي قدمتها ميسترال سابا இந்த கட்டுரை முதலில் ஆங்கிலத்
A new method translates programming logic into natural language to boost problem-solving flexibility. Large language models have become adept at narrow tasks like solving math problems or writing code snippets. But when it comes to flexible, cross-domain reasoning—connecting logical dots between scientific concepts or untangling multi-step real-world puzzles—their
Large language models can draft emails, summarize meetings, and even tell a decent joke. But ask one to untangle a thorny supply chain problem or debug a complex algorithm, and it might flounder—or worse, confidently spit out a plausible-sounding fiction. A new study reveals how two key upgrades—structured