Open
Description
diffusers 0.28.0.dev0
first time runs well, the second runtime changed the content scale to 2 , CUDA out of memory errors.
- guess the content scale will be transfered to float ?
- gradio will not release gpu after the first error
:
File "/home/notebook/code/personal/CSGO/gradio/app.py", line 172, in create_image
images = csgo.generate(pil_content_image=content_image, pil_style_image=style_image,
File "/home/notebook/code/personal/CSGO/./ip_adapter/ip_adapter.py", line 735, in generate
images = self.pipe(
File "/opt/conda/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/home/notebook/code/group/diffusers/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py", line 1235, in __call__
self.check_inputs(
File "/home/notebook/code/group/diffusers/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py", line 753, in check_inputs
raise TypeError("For single controlnet: `controlnet_conditioning_scale` must be type `float`.")
TypeError: For single controlnet: `controlnet_conditioning_scale` must be type `float`.
return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB (GPU 0; 31.75 GiB total capacity; 28.56 GiB already allocated; 8.75 MiB free; 29.92 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
PTAL
Metadata
Metadata
Assignees
Labels
No labels