Mistral AILabs & éditeursEN
Introducing Timestral (s'ouvre sur Mistral AI, nouvel onglet)
State-of-the-art probabilistic forecasting in four sizes (5M–209M params), CPU-efficient, zero-shot ready.
41 flux RSS officiels · actualisés toutes les heures
Ressources · Actualité IA
L'actualité de l'IA, puisée à la source.
Laboratoires et éditeurs, presse spécialisée, autorités de régulation : les articles récents réunis en un seul fil. Chaque article s'ouvre sur le site qui l'a publié.
1 flux sur 41 n'ont pas pu être chargés. Les autres sources restent disponibles.
- Apple Machine LearningLabs & éditeursEN
Local and Distributed Large LLM Inference with MLX and RDMA on Mac Studio (s'ouvre sur Apple Machine Learning, nouvel onglet)
MLX on Mac with Apple silicon lets AI researchers and developers experiment locally with the latest models and methods in a private environment. The open-source array framework leverages the unified memory architecture of Apple silicon, enabling anyone to run new inference and fi…
- Apple Machine LearningLabs & éditeursEN
Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents (s'ouvre sur Apple Machine Learning, nouvel onglet)
Multimodal agents commonly generate freeform reasoning before each action. For small models, limited model capacity can result in lengthy reasoning that provides little useful guidance for action generation while incurring substantial inference cost. To address this challenge, we…
- Apple Machine LearningLabs & éditeursEN
Plan or Retrieve First? A Scale Study of Tool Discovery Architectures (s'ouvre sur Apple Machine Learning, nouvel onglet)
Tool-augmented LLM agents discover relevant APIs from large registries before planning execution, yet finite context windows prohibit injecting all tool definitions into a single prompt. The dominant approach, which we term Retrieve-then-Plan (RtP), embeds the user query and retr…
- Together AILabs & éditeursEN
Meta Muse Spark Open is live on Together AI (s'ouvre sur Together AI, nouvel onglet)
Meta Muse Spark Open, Meta's open-weight model for agentic coding, is live on Together AI with a 1M-token context window on serverless and dedicated infrastructure.
- Google ResearchLabs & éditeursEN
Putting Earth AI to work: Geospatial agents for our planet's biggest challenges (s'ouvre sur Google Research, nouvel onglet)
Earth AI
- Apple Machine LearningLabs & éditeursEN
Code-Switching Detection via Language-Selective Dimensions in Speech Foundation Models (s'ouvre sur Apple Machine Learning, nouvel onglet)
Code-switching (CS), the alternation between languages within an utterance, is common in multilingual speech but underrepresented in pretraining corpora. Increasing CS coverage requires finding relevant utterances in large speech collections so that they can be included in traini…
- OllamaLabs & éditeursEN
Run AI agents safely with Ollama and Fly.io (s'ouvre sur Ollama, nouvel onglet)
AI agents can now use Ollama's cloud models safely inside Fly.io's cloud computers.
- InterconnectsPresse & veilleEN
I expect rapid progress but not towards general superintelligence (s'ouvre sur Interconnects, nouvel onglet)
I’ve often been surprised when I hear from top researchers in industry that they think AI will be better than them at their job in a few years, and I didn’t really know why I doubted it.
- InterconnectsPresse & veilleEN
The Cyber Risk Discourse is Broken (s'ouvre sur Interconnects, nouvel onglet)
Open-weights, ideology, and acknowledging trade-offs.
- Together AILabs & éditeursEN
Expanding our enterprise inference capacity with IBM Cloud and NVIDIA (s'ouvre sur Together AI, nouvel onglet)
Enterprises can now run open models at production scale on a dedicated B300 inference cluster, built by Together AI, IBM Cloud, and NVIDIA
