[Bug] AttributeError: 'SVDQW4A4Linear' object has no attribute 'weight_function'

[ThuGie]

Checklist

• 1. I have searched for related issues and FAQs (https://github.com/nunchaku-tech/nunchaku/blob/main/docs/faq.md) but was unable to find a solution.
• 2. The issue persists in the latest version.
• 3. Please note that without environment information and a minimal reproducible example, it will be difficult for us to reproduce and address the issue, which may delay our response.
• 4. If your report is a question rather than a bug, please submit it as a discussion at https://github.com/nunchaku-tech/ComfyUI-nunchaku/discussions/new/choose. Otherwise, this issue will be closed.
• 5. I will do my best to describe the issue in English.

Describe the Bug

using it for z-image turbo, without lora it works fine the moment i add a lora doesnt matter what lora i get this issue..

Environment

pytorch version: 2.9.1+cu130
Python version: 3.12.10 (tags/v3.12.10:0cc8128, Apr 8 2025, 12:21:36) [MSC v.1943 64 bit (AMD64)]
ComfyUI version: 0.12.0
ComfyUI frontend version: 1.37.11

Reproduction Steps

load lora provided workflow for z-image turbo. set a lora any lora to be honest, and just start a prompt.

[ssitu]

Here is a simple fix. Go to ComfyUI\custom_nodes\ComfyUI-nunchaku\model_patcher\zimage.py and go to line 217.
Replace this

                                else:

with this

                                elif hasattr(m, "weight_function"):

So the full snippet should look like this:

                        if lowvram_possible:
                            if weight_key in self.patches:
                                if force_patch_weights:
                                    self.patch_weight_to_device(weight_key)
                                elif hasattr(m, "weight_function"):
                                    _, set_func, convert_func = get_key_weight(self.model, weight_key)
                                    m.weight_function.append(
                                        LowVramPatch(weight_key, self.patches, convert_func, set_func)
                                    )
                                    patch_counter += 1

However, there seems to be crashing when switching LoRAs in a certain way; not sure if it is because of this change or something else.
Edit: the crashing seems to be OOMs that are unrelated, so this is probably fine.

[pedroa75]

I was having this issue for 2 months.

This fixes it for me, using SwarmUI with ComfyUI v0.14.0, nunchaku v1.2.1.

Thanks @ssitu

[sandichhuu]

I got the same error.
Any update ?

[Froztby]

That fix doesn't seem to work anymore.

[sandichhuu]

I found the solution.
Just remove the --fast flag on comfyui run command.

[dxyderderder123]

Nunchaku-Z-Image-Turbo-LoRA-修复指南.md

[shermanmxf]

Nunchaku-Z-Image-Turbo-LoRA-修复指南.md

that works for me