Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

using multistep schedulers with panorama causes an error #465

Open
ssube opened this issue Jan 15, 2024 · 0 comments
Open

using multistep schedulers with panorama causes an error #465

ssube opened this issue Jan 15, 2024 · 0 comments
Labels
pipeline/panorama scope/api status/confirmed issues that have been discussed but not planned type/bug broken features

Comments

@ssube
Copy link
Owner

ssube commented Jan 15, 2024

Using any of the multisteps schedulers with the panorama pipeline eventually causes an error where the scheduler runs out of steps:

Traceback (most recent call last):
  File "/opt/onnx-web/api/onnx_web/chain/pipeline.py", line 244, in __call__
    stage_result = stage_pipe.run(                    
  File "/opt/onnx-web/api/onnx_web/chain/source_txt2img.py", line 144, in run
    output = pipe(                                                                                                                                                                                                                                                           
  File "/opt/onnx-web/api/onnx_web/diffusers/pipelines/panorama_xl.py", line 953, in __call__                                                                                                                                                                                
    return StableDiffusionXLPanoramaPipelineMixin.__call__(self, *args, **kwargs)                                                                                                                                                                                            
  File "/opt/onnx-web/api/onnx_web/diffusers/pipelines/panorama_xl.py", line 946, in __call__                                                                                                                                                                                
    return self.text2img(*args, **kwargs)                                                                                                                                                                                                                                    
  File "/opt/onnx-web/api/onnx_web/diffusers/pipelines/panorama_xl.py", line 458, in text2img                                                                                                                                                                                
    scheduler_output = self.scheduler.step(                                                                                                                                                                                                                                  
  File "/opt/onnx-web/api/onnx_env/lib/python3.10/site-packages/diffusers/schedulers/scheduling_dpmsolver_singlestep.py", line 849, in step                                                                  
    order = self.order_list[self.step_index]                                                                                                                                                                                                                                 
IndexError: list index out of range    

It looks like the multistep scheduler (and DPM singlestep) generally keep some internal state tracking the current step, which needs to be reset/frozen during panorama pipelines.

@ssube ssube added status/confirmed issues that have been discussed but not planned type/bug broken features scope/api pipeline/panorama labels Jan 15, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
pipeline/panorama scope/api status/confirmed issues that have been discussed but not planned type/bug broken features
Projects
None yet
Development

No branches or pull requests

1 participant