AI BriefingHuggingFace Blog news
The latest AI and developer news from HuggingFace Blog, with short summaries.
Visit website →8-bit Quantization for LLM Memory Reduction
It explains an 8-bit quantization technique that cuts memory usage of large language models by 2x while preserving performance, and its integration into the Hugging Face library.
Hugging Face Releases Skops
Hugging Face has released the Skops library, which supports hub hosting and documentation of scikit-learn models.
Guide to Deploying ViT on Kubernetes
It covers how to deploy a Vision Transformer model into a scalable environment using TensorFlow Serving and Kubernetes.
Hugging Face unveils Enterprise Hub
Hugging Face has launched Enterprise Hub, integrating enterprise-grade security and collaboration tools.
Nyströmformer: Efficient Approximation
We introduce Nyströmformer, which uses the Nyström method to reduce the complexity of Transformer's self-attention from $O(n^2)$ to $O(n)$.
Datasets Audio and Vision Update
New documentation and features for processing audio and image data have been added to the Hugging Face Datasets library.
100x Faster Text Generation with TensorFlow XLA
By combining TensorFlow and XLA in Hugging Face's `transformers`, text generation speed can be increased by up to 100x.
A Guide to Deploying Vision Models with TF Serving
This explains how to deploy Hugging Face's TensorFlow vision models using TF Serving.
Analysis of BLOOM Model Training Infrastructure
This details the hardware and software technology used to train the 176B-parameter BLOOM model.
176B-Parameter Open Model BLOOM Released
BLOOM, the world's largest open-source multilingual language model with 176 billion parameters, has been released.
Accelerating Large-Scale Model Training with DeepSpeed
It introduces how to apply DeepSpeed ZeRO technology through Hugging Face's Accelerate library to solve memory issues in large-scale model training.
Transformers ONNX Conversion Guide
Explains how to convert Transformers models to ONNX format and optimize inference performance using Hugging Face Optimum.
Intel-Hugging Face, Model Acceleration Collaboration
Intel and Hugging Face have partnered to release the Optimum Intel library, which supports hardware acceleration for Transformer models.
Mathematical Principles of the DDPM Diffusion Model and PyTorch Implementation
This article explains the mathematical principles of DDPM in detail and provides step-by-step implementation code using PyTorch.
Graphcore Releases Transformers for IPU
A variety of Transformer models optimized for Graphcore IPUs have been added to the Hugging Face Optimum library.
Hugging Face Launches Collaboration Features
Pull Request and Discussion features have been added to the Hugging Face Hub for models, datasets, and Spaces.
TAPEX: Table Pre-training via SQL Execution
Introducing TAPEX, which efficiently trains table tasks using synthetic data derived from SQL execution results.
Gradio 3.0 Released: Introducing Blocks
Gradio has released version 3.0, which includes a frontend redesign and the Blocks API for custom layouts.
Hugging Face Supports Inference Acceleration for Optimum
Hugging Face has updated the features of the Optimum library, which optimizes inference performance for Transformers models.
Hugging Face Raises $100 Million Series C Funding
Hugging Face has raised a $100 million Series C round led by Lux Capital.