This is the full packaged SD3.5 with a full epoch of about 915k images.
Package Contains:
SD3.5 M full
Frozen Omega-FP32 CLIP_L
Frozen Omega-FP32 CLIP_G
950k samples trained T5xxl-Unchained
This will likely NOT WORK CORRECTLY until I release my ComfyUI Clip loader FP Q converter pipeline.
Description
The train was a fair success. It's responding to grid and similar grid-based options using the T5.