AI BriefingHuggingFace Blog news
The latest AI and developer news from HuggingFace Blog, with short summaries.
Visit website →LLM Performance Optimization: Understanding Prefill/Decode
This explains the mechanisms and key performance metrics of Prefill and Decode, the core stages of LLM inference.
LCLM Evaluation Benchmark HELMET Released
Introducing HELMET, a new benchmark that evaluates Long-context language models from multiple angles, going beyond existing simple synthetic tasks.
Cohere Now Available for Direct Inference on Hugging Face
Cohere's major LLM and multimodal models are now available for serverless inference on the Hugging Face Hub.
Gradio, an ML Framework Beyond Simple UI
Gradio is a framework for interacting with ML models, featuring automatic API generation, SSR, and queue management.
Hugging Face Model Security Scanning Achievements
Hugging Face and Protect AI collaborated to scan 4.47 million models and detect over 350,000 security threats.
Hugging Face Acquires Pollen Robotics
Hugging Face is entering the open-source robotics market in earnest by acquiring Pollen Robotics.
Multimodal Model Visual Salamandra Released
Visual Salamandra, a 7B-parameter multimodal model, has been released with image and video understanding capabilities.
Arabic LLM Evaluation Leaderboard and AraGen Update
A unified leaderboard for evaluating Arabic LLM performance has launched, along with updates to the AraGen and Arabic IFEval benchmarks.
Llama 4 Maverick and Scout Unveiled
Meta's next-generation MoE models, Llama 4 Maverick and Scout, have been released on Hugging Face.
Hugging Face Overhauls LLM Course
Hugging Face is updating and overhauling its existing NLP course into an LLM-focused, up-to-date curriculum.
Request Queuing Strategies for LLM Performance Optimization
This proposes introducing fair scheduling at the API server layer to prevent specific users from monopolizing requests.
Hugging Face Strengthens AI Infrastructure Security with Infisical
Hugging Face adopted Infisical to automate secrets management and strengthen security across its multi-cloud environment.
TGI Begins Native Support for Intel Gaudi
HuggingFace's LLM serving solution TGI now natively supports Intel Gaudi hardware.
DeepSeek-V3-0324 Model Released
DeepSeek has released DeepSeek-V3-0324, a model with improved performance and licensing.
Sentence Transformers v4.0 Adds Reranker Training Support
The Sentence Transformers v4.0 update reveals an efficient way to train and fine-tune Reranker models.
Gradio Announces Major Update to Dataframe Component
Gradio has announced an update to the Dataframe component that greatly enhances data manipulation and visualization features.
Hugging Face Updates Real-Time Analytics Feature for Inference Endpoints
Hugging Face announced a real-time analytics dashboard update for monitoring and debugging Inference Endpoints.
Hugging Face Emphasizes Open Source to the White House
Hugging Face submitted a response to the White House's AI Action Plan emphasizing the core role of open source and open science.
Hugging Face Adopts Xet-Based Storage
Hugging Face has introduced Xet storage, which supports byte-level deduplication, to improve the efficiency of large file transfers.
NVIDIA Unveils Physical AI Models and Datasets
NVIDIA has released the Cosmos Transfer model, a large-scale dataset, and Isaac GR00T N1 to advance robotics and autonomous driving.