AI BriefingHuggingFace Blog news
The latest AI and developer news from HuggingFace Blog, with short summaries.
Visit website →The MoE Structure and How It Works in Transformers
This explains the principles and advantages of the MoE (Mixture of Experts) structure, which maximizes the computational efficiency of Transformers.
Training LLMs with Unsloth and HF Jobs
Introduces how to efficiently fine-tune LLMs using coding agents by combining Unsloth with Hugging Face Jobs.
llama.cpp joins Hugging Face
GGML, the team behind llama.cpp, has joined Hugging Face to strengthen support for the local AI ecosystem.
IBM, UC Berkeley Identify Root Causes of AI Agent Failures
IBM and UC Berkeley have released MAST, a methodology that systematically analyzes failure patterns in IT automation agents.
Gradio 6, Custom HTML/CSS/JS Support
Gradio 6's `gr.HTML` update now enables implementation of complex interactive web components within a single Python file.
New Technique Unveiled for Writing CUDA Kernels for AI Agents
HuggingFace has unveiled a new skill that helps AI agents write CUDA kernels directly.
Meta and Hugging Face Release OpenEnv for Agent Evaluation
Meta and Hugging Face have released OpenEnv, a framework for evaluating AI agents in real-world environments.
Transformers.js v4 Released
Transformers.js v4, featuring the introduction of a WebGPU runtime and performance optimizations, has been released via NPM.
SyGra Studio: A Visual Synthetic Data Generation Tool
ServiceNow has unveiled SyGra Studio, which lets users design synthetic data generation workflows through a visual interface.
Hugging Face Launches Decentralized Evaluation System
Hugging Face has launched a decentralized evaluation system that allows the community to directly submit and verify scores, aiming to increase transparency in model evaluation.
H Company Unveils UI-Specialized Holo2 Model
H Company has unveiled a new Holo2-235B-A22B model that maximizes UI element localization performance.
Chinese AI Leads the Global Open-Source Ecosystem
Chinese AI companies are emerging as the central axis of the open-source ecosystem by actively releasing their models and research achievements.
T2I Model Training Optimization Guide
Photoroom has released experimental analysis and optimization techniques for efficient Text-to-Image model training.
Daggr: AI Workflow Visualization Tool Released
Daggr, developed by the Gradio team, is an open-source library that lets you define AI workflows as code and visually debug them.
Agent Skill-Transfer Tool 'upskill' Released
Introducing 'upskill', a tool that transfers the capabilities of high-performance models to smaller models, helping them perform complex tasks.
Three Major Technology Trends in China's Open-Source AI
China's open-source AI ecosystem is evolving through MoE architecture, multimodal expansion, and small model optimization.
Emirati Dialect Evaluation Benchmark Alyah Released
A new benchmark called Alyah has been released to measure Arabic LLMs' understanding of the Emirati dialect.
GPT-OSS Agentic RL Training Methodology Revealed
It covers the technical challenges and solutions involved in applying Agentic RL to the GPT-OSS model to optimize agent performance.
IBM Unveils Benchmark for Industrial AI Agents
AssetOpsBench, an AI agent benchmark for evaluating complex operational environments in industrial settings, has been released.
How DeepSeek R1 Transformed the Open-Source AI Ecosystem
Since the release of DeepSeek R1, the China-centered open-source AI ecosystem has grown rapidly, reshaping the landscape of the global market.