Skip to content

arnavdantuluri/StableTriton

Repository files navigation

StableTriton (WIP) (Diffusers implementation working; Boosts it/s from 5.94it/s to 8.58it/s with triton and cuda graphs)

A Triton inference engine built in the same grain as Kernl https://github.com/ELS-RD/kernl/ but for diffusion models, specifically for SDXL

Usage

When done you will simply need to run model = compile(model). As long as it is a nn.Module torch.fx should overwrite native torch ops with triton ops. Will initally set it up with ComfyUI and Diffusers. Automatic1111 will be worked on soon after

About

The first open source triton inference engine for Stable Diffusion, specifically for sdxl

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published