ModelVerificadoExterno

Qwen Image Lightning

Please refer to Qwen-Image-Lightning github to learn how to use the models.

Fuentehuggingface.co/lightx2v/Qwen-Image-Lightning/tree/e74da8d4e71a54b341de86aa9f8d2509165aa513

TipoOtherModelo baseQwen/Qwen-ImageVersiónRevision e74da8dPublicador@lightx2vBcalidad

Fuente del modelo

Tipo: Other
Tamaño del archivo: 810 MB–19,0 GB
Modelo base: Qwen/Qwen-Image
Versión: vRevision e74da8d
Actualizado: 5 oct 2026
Publicador: @lightx2v
Descargas: 331,8 mil
Tareas: Edición de imagen, Texto a imagen
Fuente: Hugging Face

Descripción de la fuente

Please refer to Qwen-Image-Lightning github to learn how to use the models.

use with diffusers 🧨:

make sure to install diffusers from main (pip install git+https://github.com/huggingface/diffusers.git)

from diffusers import DiffusionPipeline, FlowMatchEulerDiscreteScheduler
import torch 
import math

# From https://github.com/ModelTC/Qwen-Image-Lightning/blob/342260e8f5468d2f24d084ce04f55e101007118b/generate_with_diffusers.py#L82C9-L97C10
scheduler_config = {
    "base_image_seq_len": 256,
    "base_shift": math.log(3),  # We use shift=3 in distillation
    "invert_sigmas": False,
    "max_image_seq_len": 8192,
    "max_shift": math.log(3),  # We use shift=3 in distillation
    "num_train_timesteps": 1000,
    "shift": 1.0,
    "shift_terminal": None,  # set shift_terminal to None
    "stochastic_sampling": False,
    "time_shift_type": "exponential",
    "use_beta_sigmas": False,
    "use_dynamic_shifting": True,
    "use_exponential_sigmas": False,
    "use_karras_sigmas": False,
}
scheduler = FlowMatchEulerDiscreteScheduler.from_config(scheduler_config)
pipe = DiffusionPipeline.from_pretrained(
    "Qwen/Qwen-Image", scheduler=scheduler, torch_dtype=torch.bfloat16
).to("cuda")
pipe.load_lora_weights(
    "lightx2v/Qwen-Image-Lightning", weight_name="Qwen-Image-Lightning-8steps-V1.0.safetensors"
)

prompt = "a tiny astronaut hatching from an egg on the moon, Ultra HD, 4K, cinematic composition."
negative_prompt = " "
image = pipe(
    prompt=prompt,
    negative_prompt=negative_prompt,
    width=1024,
    height=1024,
    num_inference_steps=8,
    true_cfg_scale=1.0,
    generator=torch.manual_seed(0),
).images[0]
image.save("qwen_fewsteps.png")

Recursos relacionados

40 conexiones

Fuentes

1 fuente

Hugging FacePrincipalVersión e74da8d4e71a54b341de86aa9f8d2509165aa513

huggingface.co/lightx2v/Qwen-Image-Lightning/tree/e74da8d4e71a54b341de86aa9f8d2509165aa513

Verificado 5 oct

Artefactos del modelo

21 artefactos

Qwen-Image-Edit-2509/qwen_image_edit_2509_fp8_e4m3fn_scaled.safetensors

safetensors · 19,0 GB · SHA-256 cc7a73527bd4…a001 · Hugging Face

Descargar

Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16.safetensors

safetensors · 810 MB · SHA-256 2a32ce938ec7…1377 · Hugging Face

Descargar

Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-4steps-V1.0-fp32.safetensors

safetensors · 1,58 GB · SHA-256 6065f62669c9…36b1 · Hugging Face

Descargar

Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-8steps-V1.0-bf16.safetensors

safetensors · 810 MB · SHA-256 b5c48ba90d22…7f72 · Hugging Face

Descargar

Qwen-Image-Edit-2509/Qwen-Image-Edit-2509-Lightning-8steps-V1.0-fp32.safetensors

safetensors · 1,58 GB · SHA-256 eece790fc6ac…e841 · Hugging Face

Descargar

Qwen-Image-Edit-Lightning-4steps-V1.0-bf16.safetensors

safetensors · 810 MB · SHA-256 d8132c32e7df…beb1 · Hugging Face

Descargar

Qwen-Image-Edit-Lightning-4steps-V1.0.safetensors

safetensors · 1,58 GB · SHA-256 e752ea1553f0…cc1b · Hugging Face

Descargar

Qwen-Image-Edit-Lightning-8steps-V1.0-bf16.safetensors

safetensors · 810 MB · SHA-256 36ee4ff86085…01b9 · Hugging Face

Descargar

Qwen-Image-Edit-Lightning-8steps-V1.0.safetensors

safetensors · 1,58 GB · SHA-256 5910104f8922…a826 · Hugging Face

Descargar

Qwen-Image-fp8-e4m3fn-Lightning-4steps-V1.0-bf16.safetensors

safetensors · 810 MB · SHA-256 11704ff6034a…442c · Hugging Face

Descargar

Qwen-Image-fp8-e4m3fn-Lightning-4steps-V1.0-fp32.safetensors

safetensors · 1,58 GB · SHA-256 a74d58821ed0…ec6b · Hugging Face

Descargar

Qwen-Image-Lightning-4steps-V1.0-bf16.safetensors

safetensors · 810 MB · SHA-256 3ac242b4ff22…8c84 · Hugging Face

Descargar

Qwen-Image-Lightning-4steps-V1.0.safetensors

safetensors · 1,58 GB · SHA-256 9526e90d71c4…c8b1 · Hugging Face

Descargar

Qwen-Image-Lightning-4steps-V2.0-bf16.safetensors

safetensors · 810 MB · SHA-256 3e43b9796143…67c5 · Hugging Face

Descargar

Qwen-Image-Lightning-4steps-V2.0.safetensors

safetensors · 1,58 GB · SHA-256 878c519b75aa…c228 · Hugging Face

Descargar

Qwen-Image-Lightning-8steps-V1.0.safetensors

safetensors · 1,58 GB · SHA-256 07b5a9998814…fff9 · Hugging Face

Descargar

Qwen-Image-Lightning-8steps-V1.1-bf16.safetensors

safetensors · 810 MB · SHA-256 fe04c3ec1dd4…79c8 · Hugging Face

Descargar

Qwen-Image-Lightning-8steps-V1.1.safetensors

safetensors · 1,58 GB · SHA-256 c5f33b60c0e3…a508 · Hugging Face

Descargar

Qwen-Image-Lightning-8steps-V2.0-bf16.safetensors

safetensors · 810 MB · SHA-256 5bdbf6991228…57ee · Hugging Face

Descargar

Qwen-Image-Lightning-8steps-V2.0.safetensors

safetensors · 1,58 GB · SHA-256 f4f5b6607939…3d56 · Hugging Face

Descargar

Qwen-Image/qwen_image_fp8_e4m3fn_scaled.safetensors

safetensors · 19,0 GB · SHA-256 0b4ecbb63b40…7bae · Hugging Face

Descargar

Extractos de fuentes

3 extractos

Readme11 julhuggingface.co/lightx2v/Qwen-Image-Lightning

--- license: apache-2.0 language: - en - zh base_model: - Qwen/Qwen-Image pipeline_tag: text-to-image tags: - Qwen-Image - distillation - LoRA - lora library_name: diffusers --- Please refer to [Qwen-Image-Lightning github](https://github.com/ModelTC/Qwen-Image-Lightning/) to learn how to use the models. use with diffusers 🧨: make sure to install diffusers from `main` (`pip install git+https://github.com/huggingface/diffusers.git`) ``` from diffusers import DiffusionPipeline, FlowMatchEulerDiscreteScheduler import torch import math # From https://github.com/ModelTC/Qwen-Image-Lightning/blob/342260e8f5468d2f24d084ce04f55e101007118b/generate_with_diffusers.py#L82C9-L97C10 scheduler_config = { "base_image_seq_len": 256, "base_shift": math.log(3), # We use shift=3 in distillation "invert_sigmas": False, "max_image_seq_len": 8192, "max_shift": math.log(3), # We use shift=3 in distillation "num_train_timesteps": 1000, "shift": 1.0, "shift_terminal": None, # set shift_terminal to None "stochastic_sampling": False, "time_shift_type": "exponential", "use_beta_sigmas": False, "use_dynamic_shifting": True, "use_exponential_sigmas": False, "use_karras_sigmas": False, } scheduler = FlowMatchEulerDiscreteScheduler.from_config(scheduler_config) pipe = DiffusionPipeline.from_pretrained( "Qwen/Qwen-Image", scheduler=scheduler, torch_dtype=torch.bfloat16 ).to("cuda") pipe.load_lora_weights( "lightx2v/Qwen-Image-Lightning", weight_name="Qwen-Image-Lightning-8steps-V1.0.safetensors" ) prompt = "a tiny astronaut hatching from an egg on the moon, Ultra HD, 4K, cinematic composition." negative_prompt = " " image = pipe( prompt=prompt, negative_prompt=negative_prompt, width=1024, height=1024, num_inference_steps=8, true_cfg_scale=1.0, generator=torch.manual_seed(0), ).images[0] image.save("qwen_fewsteps.png") ```

Note5 octhuggingface.co/lightx2v/Qwen-Image-Lightning

Source context: 331759 downloads · 835 likes · Pipeline text-to-image · Library diffusers · Repo lightx2v/Qwen-Image-Lightning

Note20 julhuggingface.co/lightx2v/Qwen-Image-Lightning

Source context: 396109 downloads · 811 likes · Pipeline text-to-image · Library diffusers · Repo lightx2v/Qwen-Image-Lightning