---
title: comfyui_nunchaku_lora_patch
canonical_url: "https://www.modelscope.cn/models/AI-ModelScope/comfyui_nunchaku_lora_patch"
md_url: "https://www.modelscope.cn/models/AI-ModelScope/comfyui_nunchaku_lora_patch.md"
repository: AI-ModelScope/comfyui_nunchaku_lora_patch
last_updated: 2025-07-28
license: apache-2.0
parameters: 230.0M
tensor_type:
  - F32
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 264
stars: 0
---

# comfyui_nunchaku_lora_patch

> comfyui_nunchaku_lora_patch - AI-ModelScope 在 ModelScope 开源的模型。Official Fix (2025-07-28) The official fix is being tested: https://github.com/nunchaku-tech/nunchaku/pull/557

AI-ModelScope/comfyui_nunchaku_lora_patch 是 ModelScope 魔搭社区上的 230.0M 参数机器学习模型，采用 apache-2.0 许可。

- **Repository**: AI-ModelScope/comfyui_nunchaku_lora_patch
- **License**: apache-2.0
- **Parameters**: 230.0M
- **Downloads**: 264
- **Stars**: 0
- **Last updated**: 2025-07-28

Source: https://www.modelscope.cn/models/AI-ModelScope/comfyui_nunchaku_lora_patch

---

## Official Fix (2025-07-28)
The official fix is being tested: https://github.com/nunchaku-tech/nunchaku/pull/557

--

## ✅ Patched LoRAs

| Patched LoRAs   | Original LoRAs |
|-----------------|----------------|
| [flux_kontext_deblur_nunchaku.safetensors](https://huggingface.co/lym00/comfyui_nunchaku_lora_patch/blob/main/flux_kontext_deblur_nunchaku.safetensors) | [civitai](https://civitai.com/models/1737381) or [civitaiarchive](https://civitaiarchive.com/models/1737381) |
| [flux_kontext_face_detailer_nunchaku.safetensors](https://huggingface.co/lym00/comfyui_nunchaku_lora_patch/blob/main/flux_kontext_face_detailer_nunchaku.safetensors) | [civitai](https://civitai.com/models/1752776) or [civitaiarchive](https://civitaiarchive.com/models/1752776) |

## ⚡ Usage

- These patched LoRAs are **compatible with** [ComfyUI-nunchaku](https://github.com/mit-han-lab/ComfyUI-nunchaku).
- Use the **Nunchaku FLUX LoRA Loader** node to load LoRA modules for **SVDQuant FLUX** models.

## 🛠️ Patch References

Some original FLUX LoRA files were missing required `final_layer.adaLN` weights needed by **ComfyUI-nunchaku’s FLUX LoRA Loader**.  
This patch script automatically adds **dummy adaLN tensors** to make the LoRA compatible with **SVDQuant FLUX** models.

**Script:** [patch_comfyui_nunchaku_lora.py](https://huggingface.co/lym00/comfyui_nunchaku_lora_patch/blob/main/patch_comfyui_nunchaku_lora.py)

**Based on:**
- **Nunchaku Issue:** [ComfyUI-nunchaku #340](https://github.com/mit-han-lab/ComfyUI-nunchaku/issues/340)  
  > Node Type: `NunchakuFluxLoraLoader`  
  > Exception Type: `KeyError`  
  > Exception Message: `'lora_unet_final_layer_adaLN_modulation_1.lora_down.weight'`

- **Example Gist:** [akedia/e0a132b5...](https://gist.github.com/akedia/e0a132b587e30413665d299ad893a60e)

---

## 🔄 Example Patch Log

```bash
Running patch_comfyui_nunchaku_lora.py

🔄 Universal final_layer.adaLN LoRA patcher (.safetensors)

Enter path to input LoRA .safetensors file: Flux_kontext_deblur.safetensors
Enter path to save patched LoRA .safetensors file: flux_kontext_deblur_nunchaku.safetensors

✅ Loaded 610 tensors from: Flux_kontext_deblur.safetensors

🔑 Found final_layer-related keys:
  - lora_unet_final_layer_linear.lora_down.weight
  - lora_unet_final_layer_linear.lora_up.weight

🔍 Checking for final_layer keys with prefix 'lora_unet_final_layer'
  Linear down: lora_unet_final_layer_linear.lora_down.weight
  Linear up:   lora_unet_final_layer_linear.lora_up.weight
  ✅ Has final_layer.linear: True
  ✅ Has final_layer.adaLN_modulation_1: False

✅ Added dummy adaLN weights:
  - lora_unet_final_layer_adaLN_modulation_1.lora_down.weight (shape: torch.Size([16, 3072]))
  - lora_unet_final_layer_adaLN_modulation_1.lora_up.weight   (shape: torch.Size([64, 16]))

✅ Patched file saved to: flux_kontext_deblur_nunchaku.safetensors
   Total tensors now: 612

🔍 Verifying patched keys:
  - lora_unet_final_layer_adaLN_modulation_1.lora_down.weight
  - lora_unet_final_layer_adaLN_modulation_1.lora_up.weight
  - lora_unet_final_layer_linear.lora_down.weight
  - lora_unet_final_layer_linear.lora_up.weight

✅ Contains adaLN after patch: True