Overview
The Chat UI is a chat interface that utilizes open-source models such as OpenAssistant or Llama. It is built on SvelteKit and powers the HuggingChat app on hf.co/chat. This repository provides a fast deploy alternative for those who do not want to configure and launch their own Chat UI.
Features
- Fast deploy alternative for Chat UI setup and launch
- Customizable Chat UI instances with supported LLMs on Hugging Face Spaces
- Compatibility with HF Space secrets for deploying models with gated access or from private repositories
- Integration with Inference for PROs curated list of powerful models with higher rate limits
- Web search feature that generates search queries from user prompts, performs web search and extracts content, and uses embeddings for vector similarity search
- Optional OpenID authentication for users
- Customizable theming using environment variables
Installation
To install and run the Chat UI locally, follow these steps:
- Create a .env.local file in the root of the repository.
- Override the default config values in the .env file with your local configuration in .env.local.
- Set up a MongoDB instance for storing the chat history. You can either use a local MongoDB instance or a free MongoDB Atlas instance.
- If you are using a remote inference endpoint, obtain a Hugging Face access token from your Hugging Face profile.
- If you want to enable OpenID authentication, add the required variables to your .env.local file.
- Customize the look and feel of chat-ui using the available environment variables.
Summary
The Chat UI is a versatile chat interface built on SvelteKit. It allows for easy deployment and customization, and includes features such as web search, authentication options, and theming. By utilizing open-source models and integrating with Hugging Face’s powerful inference capabilities, the Chat UI provides a seamless chat experience for users.
