GitLab

Update LoadClip to load clip models for SDXL refiner.

Tomesd now uses q instead of x to decide which tokens to merge because
it seems to give better results.

Large refactor of the model detection and loading code.

Change the transformer patches function format to be more future proof.

Make the CLIP model work on the GPU.

Default pytorch Linear initializes the weights which is useless and slow.

This means arguments can be added without issues.