avernus

This is a ml inference server based on huggingface libraries which provides a REST based API for client interactions.

Endpoints are documented by accessing the server at http://127.0.0.1:6969/docs if using the default port.

There is a basic python client available in modules/client.py

Features:

• LLM chat support.
• SDXL support for multiple models, image2image, controlnets, and ipadapters.
• Flux Support. Only dev is currently supported. image2image and ipadapters are supported.
• Qwen Image, Qwen Image Edit, and Qwen Image Edit Plus
• Wan T2V and I2V
• ACE Step
• Chroma
• HiDream
• Hunyuan Video
• Auraflow
• Lumina 2
• Upscalers
• SD 1.5
• Sana Sprint

Install:

Run the install script for either windows or linux depending on your operating system.

Start via start_avernus.sh or start_avernus.bat depending on your operating system. The startup script also contains the url and port of the avernus server, if you want to change the port or bind to a specific device.

TODO:

• Add svg model endpoint
• Add LLM chat templates
• 3d model model support