nunchaku

nunchaku copied to clipboard

Compatible with comfyUI runpod worker ?

1

Hi there, anyone successfully setup nunchaku with the docker file of the comfyUI worker ( runpod ) ? https://github.com/blib-la/runpod-worker-comfy/tree/main#use-the-docker-image-on-runpod

BenDes21

Do you think nunchaku could support Flex-Alpha in the future?

3

Since Flex-Alpha is a de-distilled version of the Flux Schnell, maybe it could be supported. I think the original unet is pruned, but maybe many things remain unchanged. Since it...

vgabbo

是不是不支持批次多图？

1

批次超过1，采样器就报错。

snowfrost

[Feature] use ComfyUI-ppm error ( ComfyUI-ppm a nice way to give flux negative)

3

use ComfyUI-ppm error ( ComfyUI-ppm a nice way to give flux negative),   GPU 0 (NVIDIA GeForce RTX 3060 Laptop GPU) 显存: 12287.375 MB 用户启用 CPU offload [2025-03-14 07:42:39.096]...

xueqing0622

Hey there! I've got a super simple question after doing a ton of code searching. Which code can show that the activation's quantization is based on a per group size...

unbelievable3513

Is it not compatible with most models in comfyui?

5

Running most workflows will report an error.

dzy1128

Hello, I am doing an experiment about lora-like layer's performance under different precision and hardware platform. I would appreciate it if you can provide a single fused kernel mentioned in...

Ther-nullptr

Please support Chroma is a 8.9 billion parameter rectified flow transformer capable of generating images from text descriptions. Based on FLUX.1 [schnell] with heavy architectural modifications.

15

https://huggingface.co/lodestones/Chroma

nitinmukesh

[Feature] quantize the customized model myself？？

3

Can I quantify the desired model myself, e.g.majicflus

lestersssss

[Feature] Support SANA 1.5

1

do you have any plan to add sana 1.5 model? or can i get a script for converting those model? https://huggingface.co/Efficient-Large-Model/SANA1.5_4.8B_1024px_diffusers

circuluspibo