Spaces:
Running
on
Zero
Apply for community grant: Company project (gpu)
Hello! First of all, thank you for your efforts and outstanding contributions to the open-source community.
This project is the official demo of the Step1X-Edit model (https://github.com/stepfun-ai/Step1X-Edit). We hope to utilize your resources to make the model more accessible and available for trial by a wider audience. Therefore, we would like to apply for access to ZeroGPU. Thank you very much!
Hi @listen2you , we've assigned ZeroGPU to this Space. Please check the compatibility and usage sections of this page so your Space can run on ZeroGPU.
@hysts Thank you very much for your assistance. I am currently working on resolving some technical issues.
I noticed in the documentation that ZeroGPU is currently using A100 GPUs with 40GB of memory. Is this still the case? If the memory usage exceeds this value, will it be impossible to use ZeroGPU?
I noticed in the documentation that ZeroGPU is currently using A100 GPUs with 40GB of memory. Is this still the case?
Actually, we are in the process of migrating underlying hardware of ZeroGPU Spaces from half A100 to one-third H200 and half H200. (Here, "half" refers to MIG configuration.)
One-third H200 will eventually be the default, but for now, to avoid breaking too many ZeroGPU Spaces at the same time, we are using half H200.
If the memory usage exceeds this value, will it be impossible to use ZeroGPU?
Yes, that's correct. CUDA OOM will occur if the memory usage exceeds the limit. But ZeroGPU Spaces are currently using half H200, so the VRAM limit is 71GB.
Hi @listen2you , congratulations on the release and on securing the ZeroGPU grants! We would love to promote the project on our Social Media channels. However, I believe the app is currently experiencing issues. I couldn't generate an edited image, as shown in my attached example. The error displayed in the logs is:
Traceback (most recent call last):
File "/usr/local/lib/python3.10/site-packages/spaces/zero/wrappers.py", line 256, in thread_wrapper
res = future.result()
File "/usr/local/lib/python3.10/concurrent/futures/_base.py", line 451, in result
return self.__get_result()
File "/usr/local/lib/python3.10/concurrent/futures/_base.py", line 403, in __get_result
raise self._exception
File "/usr/local/lib/python3.10/concurrent/futures/thread.py", line 58, in run
result = self.fn(*self.args, **self.kwargs)
File "/home/user/app/app.py", line 392, in inference
image = infer_func(
File "/usr/local/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/home/user/app/app.py", line 321, in generate_image
ref_images = self.ae.encode(ref_images_raw.to(self.device) * 2 - 1)
File "/home/user/app/modules/autoencoder.py", line 317, in encode
z = self.reg(self.encoder(x))
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "/home/user/app/modules/autoencoder.py", line 164, in forward
hs = [self.conv_in(x)]
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 554, in forward
return self._conv_forward(input, self.weight, self.bias)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 549, in _conv_forward
return F.conv2d(
RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same
Hi @listen2you , congratulations on the release and on securing the ZeroGPU grants! We would love to promote the project on our Social Media channels. However, I believe the app is currently experiencing issues. I couldn't generate an edited image, as shown in my attached example. The error displayed in the logs is:
Traceback (most recent call last): File "/usr/local/lib/python3.10/site-packages/spaces/zero/wrappers.py", line 256, in thread_wrapper res = future.result() File "/usr/local/lib/python3.10/concurrent/futures/_base.py", line 451, in result return self.__get_result() File "/usr/local/lib/python3.10/concurrent/futures/_base.py", line 403, in __get_result raise self._exception File "/usr/local/lib/python3.10/concurrent/futures/thread.py", line 58, in run result = self.fn(*self.args, **self.kwargs) File "/home/user/app/app.py", line 392, in inference image = infer_func( File "/usr/local/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context return func(*args, **kwargs) File "/home/user/app/app.py", line 321, in generate_image ref_images = self.ae.encode(ref_images_raw.to(self.device) * 2 - 1) File "/home/user/app/modules/autoencoder.py", line 317, in encode z = self.reg(self.encoder(x)) File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1762, in _call_impl return forward_call(*args, **kwargs) File "/home/user/app/modules/autoencoder.py", line 164, in forward hs = [self.conv_in(x)] File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1751, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1762, in _call_impl return forward_call(*args, **kwargs) File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 554, in forward return self._conv_forward(input, self.weight, self.bias) File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 549, in _conv_forward return F.conv2d( RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same
Hi
@ysharma
Thank you very much for your help. I really need some support here. Specifically, I found that this space will repeatedly upload incorrect examples, which have consumed all of my quota, making it impossible for me to debug this space properly. How can I resolve this issue?
Additionally, I found that this technical issue may require me to recreate a space. After recreating it, will I still have access to the zerogpu?
Here is an example for Step1X-Edit.
The app is fantastic and functioning excellently now. Thank you for addressing the issue, @listen2you ! We've amplified the project on X, LI, Bluesky, and YouTube through our accounts.
I have also opened a PR to add our new ImageSlider component to your app: https://huggingface.co./spaces/stepfun-ai/Step1X-Edit/discussions/2
I have duplicated your space and added imageslider just to demonstrate how it will look. You can access the app here: https://huggingface.co./spaces/ysharma/Step1X-Edit
Let me know your thoughts on this?
I have also opened a PR to add our new ImageSlider component to your app: https://huggingface.co./spaces/stepfun-ai/Step1X-Edit/discussions/2
I have duplicated your space and added imageslider just to demonstrate how it will look. You can access the app here: https://huggingface.co./spaces/ysharma/Step1X-Edit
Let me know your thoughts on this?
Many thanks! It is extremely fascinating, and I am merging it!
Awesome, thanks!
A big congratulations to @listen2you and the amazing StepFun team! We're thrilled to share that the Step1X-Edit app is trending this week on Huggingface !
@ysharma Thank you so much! My team and I are truly excited about this. It's an honor to contribute to the open-source community, and we're deeply grateful for your team's support. Without your help, it would have been impossible to share our work with so many community members.