LTX2.3 FP4 Models

Because no one uploads FP4 here. and they should (fp4 works on 3xxx and above for vram savings.)

5 Version's Available.
The NVFP4 is from Hippotes / LTX-2.3-various-formats
The Scaled FP4 TFO is one I made from Kijai's Scaled fp8 model. Its transformer only, so needs VAE's and Text Projection.
Distilled FP4 TFO, 19GB, Transformer Only.
Distilled FP4 Extreme @ 14.8GB, Transformer Only, load via checkpoint loader.
Dev FP4 Extreme @ 14.8gb, Transformer only, load via checkpoint loader.

Next, if i can get the output cleaned up, Split high/low models, 8.3gb each.

Should of added this awhile ago, diffusion model loader/unet loader seems to get confused with these, use the checkpoint loader even though it has no clip/vae