Downloads · 30 days
21
28% of all-time downloads
lite-infer/qwen-image-edit-2509-nunchaku-lite-int4-r32
qwen-image-edit-2509-nunchaku-lite-int4-r32 is a image-to-image model from lite-infer. Use it when you need one image transformed into another. It is set up for diffusers. The card lists the license as apache-2.0.
This directory is a Diffusers-loadable conversion of:
Downloads · 30 days
21
28% of all-time downloads
All-time downloads
74
Public
Parameters
8.3B
18.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors18.1 GB · 100%
How the weights are stored.
I86.8B · 82%
From the Hugging Face model README
This directory is a Diffusers-loadable conversion of:
Qwen/Qwen-Image-Edit-2509nunchaku-ai/nunchaku-qwen-image-edit-2509svdq-int4_r32-qwen-image-edit-2509.safetensorsThe transformer is packaged for the Diffusers Nunchaku Lite loader:
quant_method: nunchaku_liteint46432720120The text encoder is saved with BitsAndBytes 4-bit NF4:
quant_method: bitsandbytesload_in_4bit: truebnb_4bit_quant_type: nf4bnb_4bit_compute_dtype: bfloat16Measured with 40 inference steps on a 1248x832 input image. Latency is the mean of 3 measured runs after 1 warmup. Max VRAM is peak CUDA allocated memory during generation.
| Checkpoint | Latency | Max VRAM |
|---|---|---|
| Converted Diffusers Nunchaku Lite INT4 r32 + BNB4 text encoder | - | 21.20 GiB |
| Original Nunchaku INT4 r32 safetensors | - | - |
Original baseline attempted: nunchaku-ai/nunchaku-qwen-image-edit-2509/svdq-int4_r32-qwen-image-edit-2509.safetensors loaded through the native Nunchaku transformer. It failed before inference on this runtime with an AssertionError in patch_scale_key while checking checkpoint dtype compatibility.

The comparison uses the same input image, prompt, seed, and 40-step settings. The NVFP4 panel was regenerated from the current converted checkpoint.
import torch
from PIL import Image
from diffusers import QwenImageEditPlusPipeline
model_path = "lite-infer/qwen-image-edit-2509-nunchaku-lite-int4-r32"
image_path = "input.png"
output_path = "output.png"
pipe = QwenImageEditPlusPipeline.from_pretrained(model_path, torch_dtype=torch.bfloat16)
pipe.to("cuda")
image = Image.open(image_path).convert("RGB")
prompt = "Change the image to watercolor style."
result = pipe(
image=image,
prompt=prompt,
generator=torch.Generator(device="cuda").manual_seed(1),
true_cfg_scale=4.0,
negative_prompt=" ",
num_inference_steps=40,
guidance_scale=1.0,
num_images_per_prompt=1,
)
result.images[0].save(output_path)