The AI startup intends to eventually make all chat models accessible on the Hub public.
HuggingChat is a brand-new chatbot powered by AI that is currently being tested on Hugging Face. HuggingChat is capable of carrying out a variety of jobs, including as writing articles, resolving code issues, or responding to inquiries, that have recently attracted a great deal of interest to ChatGPT. According to Hugging Face, HuggingChat, which includes 30 billion parameters, is currently the best open source chat model.
HuggingChat is now based on the most recent LLaMa model created by the project OpenAssistant, as stated by Hugging Face. Making an AI-based assistant efficient and compact so that it can run on consumer hardware is another objective they are aiming towards. The LAION-5B dataset, on which Stable Diffusion is based, is one among the open datasets, tools, and models made available by LAION, a non-profit organisation that manages OpenAssistant.
HuggingChat upholds a rigorous privacy approach in which messages are only saved for user viewing and are never even shared for training or research reasons. Furthermore, cookies are not used for user authentication or identification. However, this will probably change in the future to allow users to communicate with researchers.
HuggingChat is the first open-source chat project with AI, as was previously mentioned. The UI’s source code is available on GitHub, and Hugging Face’s Inference API infrastructure is used by the inference backend to conduct text-generation-inference.
As a result, the app can be customised in a variety of ways once it is deployed to a Hugging Face Space, such as by switching out the models, changing the user interface, altering the policy on the storage of user messages, etc.