MCP App StoreLe meilleur de l'écosystème IA et MCP, sélectionné chaque jour.
Sources
Hugging Face introduces the Ettin Reranker family, designed to improve the precision of retrieval-augmented generation (RAG) pipelines. A key tool for developers looking to refine document ranking in AI applications.
IBM research demonstrates how synthetic data generation can overcome training data bottlenecks for the Granite model series. This highlights a critical shift toward curated synthetic pipelines for improving model reasoning and domain accuracy.
IBM Research introduces a new Open Agent Leaderboard on Hugging Face to provide standardized evaluation for AI agents. This helps developers benchmark agentic capabilities and research performance in open-source environments.
IBM releases Granite Embedding Multilingual R2, a high-performance multilingual embedding model under Apache 2.0. It features a 32K context window and delivers top-tier retrieval quality for models under 100M parameters.
Hugging Face explores new methods for asynchronicity in continuous batching to improve LLM inference throughput. This research aims to reduce bottlenecks in how requests are processed and served in production environments.
Amazon outlines new architectural building blocks for optimizing the training and inference of foundation models on AWS. Focuses on scalability and efficiency for large-scale AI deployments.
A showcase of a multi-agent system designed for CNC manufacturability, developed during an AMD hackathon on MI300X hardware. Demonstrates practical agent application in industrial manufacturing.
Introduction of OncoAgent, a multi-agent framework designed for oncology clinical decision support. The system focuses on privacy-preserving AI to assist in complex medical decision-making processes.
Introduction of EMO, a new approach to pretraining Mixture of Experts (MoE) that promotes emergent modularity. This research aims to improve how models specialize and efficiently allocate parameters across different tasks.
A guide on fine-tuning a clinical AI model using AMD ROCm, demonstrating a viable alternative to CUDA for medical AI development. Useful for developers looking to optimize training on AMD hardware.
A deep dive into improving RL correctness for vLLM, focusing on the transition from V0 to V1. Essential reading for those optimizing LLM serving and reinforcement learning pipelines.
Hugging Face is introducing 'Benchmaxxer Repellant' to the Open ASR Leaderboard by incorporating private test data. This move aims to combat leaderboard gaming and ensure that ASR model performance is genuine and robust.
Evaluating AI models is becoming a critical compute bottleneck as complexity increases. This shift highlights the need for more efficient evaluation frameworks to prevent a slowdown in model iteration.
Hugging Face integrates DeepInfra as an inference provider, expanding the accessibility of high-performance model hosting. This move strengthens the open-source AI ecosystem by lowering the barrier to deploying large-scale models.
NVIDIA releases Nemotron 3 Nano Omni, a multimodal model designed for long-context processing of audio, video, and documents. This enables more capable agents for complex multimodal analysis and retrieval.
Hugging Face introduces a new privacy filter designed for OpenAI web applications to help users better manage and protect their data. This tool provides an essential layer of privacy for developers and end-users interacting with AI web interfaces.
DeepSeek-V4 introduces a massive million-token context window optimized for agentic workflows. This enables agents to maintain significantly larger project contexts and long-term memory without losing retrieval accuracy.
A practical guide on integrating Transformers.js into Chrome extensions for on-device AI processing. It demonstrates how to run ML models directly in the browser, reducing latency and enhancing privacy.
NVIDIA showcases Gemma 4 VLA (Vision-Language-Action) running on Jetson Orin Nano Super. Demonstrates high-performance edge AI capabilities for robotic control and multimodal understanding in a compact form factor.
NVIDIA demonstrates how to use Nemotron personas to create synthetic demographic data for grounding AI agents in specific cultural contexts. This approach improves agent reliability and persona-consistency when deploying localized AI assistants.