Chat UI

The UI of Chat is becoming increasingly complex, often encompassing an entire front-end project along with deployment solutions.

This repository aims to construct the entire front-end UI using a single HTML file, aiming for a minimalist approach to create a chatbot.

By simplifying the structure and key functions, developers can quickly set up and experiment with a functional chatbot, adhering to a slimmed-down project design philosophy.

Features

• Supports OpenAI-format requests, enabling compatibility with various backends such as HuggingFace Text Generation Inference (TGI), vLLM, etc.

• Automatically supports multiple response formats without additional configuration, including standard OpenAI response formats, Cloudflare AI response formats, and plain text responses

• Support various backend endpoints through custom configurations, providing any project with a universal frontend chatbot

• Support the download of chat history, interrupt the current generation, and repeat the previous generation to quickly test the backend inference capabilities

• Support MCP (Model Context Protocol) by acting as a renderer and facilitating community interactions with the backend main process via IPC

• Inquiries with image inputs can be made using multimodal vision models

• Support for toggling between original format and Markdown format display

• Support internationalization and localization i18n

How to use

Option 1: Chat with demo AIQL

The demo will use Llama-3.3-70B-Instruct by default

Multimodal image upload is only supported for vision models

MCP tools call necessitates a desktop backend and LLM support in OpenAI format, referencing Chat-MCP

Option 2: Download Index and open it locally (recommended)

Option 3: Download Index and deploy it by python

cd /path/to/your/directory
python3 -m http.server 8000

Then, open your browser and access http://localhost:8000

Option 4: fork this repo and link it to Cloudflare pages

Demo: https://www2.aiql.com

Option 5: Deploy your own Chatbot by Docker

docker run -p 8080:8080 -d aiql/chat-ui

Option 6: Deploy within Huggingface

Don't forget add app_port: 8080 in README.md

Option 7: Deploy within K8s

Option 8: Integrated as a renderer within the desktop application

Demo: Chat-MCP

How to config

By default, the Chatbot will use API format as OpenAI ChatGPT.

You can insert your OpenAI API Key and change the Endpoint in configuration to use API from any other vendors

You can also download the config template from example and insert your API Key, then use it for quick configuration

Trouble Shooting

If you're experiencing issues opening the page and a simple refresh isn't resolving the issue, please take the following steps:

Reset Interface Configuration

1. Click Refresh icon on the upper right of Interface Configuration

Reset All Configuration

1. Click hidden botton on the right side of the index page
2. Click Reset All Config icon