FLUX (Base UNET) + Google FLAN FP32/NF4
This model took the 42GB FP32 Google Flan T5xxl and quantized it with improved CLIP-L for Flux. To my knowledge no one else has posted or attempted this.
Quantized from FP32 T5xxl
Base UNET no baked lora's or other changes
I noticed a speed increase with this model in addition to increased accuracy.
Note: This was done in CPU - It used 32GB of Ram and 110GB of my NVME virtual memory.