Hugging Face
7 articles found in this topic.
Hugging Face Enhances Dataset Streaming for Large-Scale AI Training
Hugging Face has significantly optimized its dataset streaming capabilities, allowing users to train AI models on terabyte-scale datasets without prior download. These enhancements boost efficiency, reduce startup requests, and accelerate data processing, even surpassing local SSD speeds in certain setups.
Zhipu AI Open-Sources AutoGLM for Automated Mobile Operations
Zhipu AI has open-sourced AutoGLM, a framework for automated, intelligent mobile operations. This release includes the core model, code, and a functional framework, operating under an MIT license. AutoGLM enables mobile devices to perform tasks through natural language commands, supporting both Chinese and multilingual applications.
Hugging Face Unveils OpenEvals for Standardized AI Model Evaluation
Hugging Face has launched OpenEvals, an open evaluation ecosystem to standardize the assessment of advanced AI models, especially LLMs. It addresses challenges like lack of standardization and reproducibility, aiming to provide a transparent and collaborative framework for the AI community.
Hugging Face Hub Library Reaches Version 1.0 After Five Years of Development
The `huggingface_hub` library has officially released version 1.0 after five years of development, marking its maturity and stability within the Python ecosystem. This milestone update introduces `httpx` as the new backend, a redesigned `hf` command-line tool, and migration to `hf_xet` for file transfers. It supports millions of models, datasets, and applications, driven by nearly 300 contributors.
Hugging Face AI Insight Talk to Feature OCR Innovations
Hugging Face's AI Insight Talk will feature advanced OCR technologies, including HunyuanOCR, PaddleOCR-VL, and MinerU. This session, in collaboration with OpenMMLab and others, will explore solutions for general recognition to multilingual support. The event is scheduled for December 4, 2025, at 20:00 Beijing Time.
PaddleOCR-VL: A Lightweight Multilingual Document Intelligence Solution for 109 Languages
PaddleOCR-VL is a new lightweight, multilingual document intelligence solution supporting 109 languages. Developed by Baidu, it integrates a visual encoder and ERNIE language model for accurate recognition of complex document elements with minimal parameters. This advancement addresses challenges in traditional OCR by offering enhanced efficiency and deployability.
Hugging Face Papers Page: Exploring Features for AI Research and Collaboration
Hugging Face has launched its Papers page, a vital resource for AI developers and researchers to track cutting-edge advancements. Curating over 10,000 high-quality papers, this platform enhances research dissemination and collaboration through features like paper claiming, community interaction, and integrated resources.