Hugging Face's logo

Spaces:
gokaygokay
/
Kolors
Running on Zero

App Files Community
6
multimodalart HF staff commited on
Commit
558526a
•
1 Parent(s): 331f30a

CPU offload worsens performance

Browse files

It saves VRAM but with ZeroGPU we don't need it

Files changed (1) hide show
  1. app.py +0 -1
app.py CHANGED
@@ -30,7 +30,6 @@ pipe = StableDiffusionXLPipeline(
30
  scheduler=scheduler,
31
  force_zeros_for_empty_prompt=False)
32
  pipe = pipe.to("cuda")
33
- pipe.enable_model_cpu_offload()
34
 
35
  @spaces.GPU(duration=200)
36
  def generate_image(prompt, negative_prompt, height, width, num_inference_steps, guidance_scale, num_images_per_prompt, use_random_seed, seed, progress=gr.Progress(track_tqdm=True)):
 
30
  scheduler=scheduler,
31
  force_zeros_for_empty_prompt=False)
32
  pipe = pipe.to("cuda")
 
33
 
34
  @spaces.GPU(duration=200)
35
  def generate_image(prompt, negative_prompt, height, width, num_inference_steps, guidance_scale, num_images_per_prompt, use_random_seed, seed, progress=gr.Progress(track_tqdm=True)):