Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
license_name: qwen
|
| 4 |
+
license_link: https://huggingface.co/Qwen/Qwen-Image-Edit-2511/blob/main/LICENSE
|
| 5 |
+
base_model: Qwen/Qwen-Image-Edit-2511
|
| 6 |
+
tags:
|
| 7 |
+
- image-editing
|
| 8 |
+
- diffusion
|
| 9 |
+
- fp8
|
| 10 |
+
- comfyui
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# Qwen-Image-Edit-2511 Transformer (FP8 e4m3fn)
|
| 14 |
+
|
| 15 |
+
FP8 quantized version of the [Qwen-Image-Edit-2511](https://huggingface.co/Qwen/Qwen-Image-Edit-2511) transformer for ComfyUI.
|
| 16 |
+
|
| 17 |
+
## Model Details
|
| 18 |
+
|
| 19 |
+
- **Original Model**: Qwen/Qwen-Image-Edit-2511
|
| 20 |
+
- **Quantization**: FP8 e4m3fn
|
| 21 |
+
- **Original Size**: 40.86 GB (BF16)
|
| 22 |
+
- **Quantized Size**: 20.43 GB
|
| 23 |
+
- **Component**: Transformer/Diffusion model only
|
| 24 |
+
|
| 25 |
+
## Usage in ComfyUI
|
| 26 |
+
|
| 27 |
+
1. Download `qwen-image-edit-2511-transformer-fp8.safetensors`
|
| 28 |
+
2. Place in `ComfyUI/models/diffusion_models/`
|
| 29 |
+
3. Use **Load Diffusion Model** node with `weight_dtype=fp8_e4m3fn`
|
| 30 |
+
4. Load text encoder separately: `qwen_2.5_vl_7b_fp8_scaled.safetensors` (or similar Qwen 2.5 VL)
|
| 31 |
+
5. Load VAE separately: `qwen_image_vae.safetensors`
|
| 32 |
+
|
| 33 |
+
## Credits
|
| 34 |
+
|
| 35 |
+
- Original model by [Qwen Team](https://huggingface.co/Qwen)
|
| 36 |
+
- FP8 conversion by armychimp
|