Hugging Face is research company that makes tools for training and using Large Language Models (LLMs), most notably the “Transformers” library widely used in working with open-source models.
The Hugging Face website is a hub for open-source models and datasets of all kinds. There are tons of models available for text generation, image generation, audio and video processing, and a whole host of other machine learning (ML) and natural language processing (NLP) tasks.
Beyond being a hub for open-source AI, Hugging Face is the center of community for people interested in running models locally on their own hardware. This is usually accomplished through the use of quantization, a technique for decreasing the computing resources needed to load and run a model. Community member Tom Jobbins a.k.a. “TheBloke” is something of a folk hero for the large number of models he has quantized for people to experiment with.
Hugging Face Links
-
Tired of AI Censorship? Open Models for Writing and Smut!
There’s a whole weird world of uncensored text generation out there, and most of it can be found on Hugging Face.
-
Wind of Change: Mistral AI’s Open-Source Models and New Mixtral 8x22B
Mistral AI has a vision to “to make frontier AI ubiquitous” and their new Mixtral 8x22B model impresses.
