New version LTX-2.3-22B-distilled-1.1
Hi! First of all, I just wanna say THANK YOU for quantizing nvfp4 of LTX2.3, I have been using LTX-2.3-22B-Distilled-FP4ME on my RTX4060 laptop with 8GB Vram+32GB Ram, it works really really well and fast (Unet loader dequantized as FP8 e4m3 + sage attention2.2). And my previous quantized version was using ltx-2.3-22b-distilled-Q5_K_M by Unsloth, it often causes VRAM Overflows for my poor 8GB Vram, it was such a depressing experience to me. So really, thanks again for making such incredible nvfp4 quantized model.
Yesterday Lightricks has uploaded the newest 1.1 version of LTX-2.3, will you consider using the newing version to make the upcoming LTX-2.3-FP4MEL ? I'll be really looking forward to you next brilliant creations.
Best regards.
+1. This also works faster than both gguf's and fp8 with amd gpu's using rocm (with bitsandbytes) It is definately faster but the existing distill fpme is just a bit below what I get from the kijai's safetensors version.