Closed harindercnvrg closed 1 year ago
Can you comment in the used versions in your environment (cloned and installed "natively" in the OS, or within a container or within a Python virtual environment)?
The requirement.txt
defines diffusers==0.2.4
, which provides PNDMScheduler
, which in deed has a add_noise()
method.
You might want to check with pip freeze > requirements.txt
, which version of diffusers
you currently have installed.
yes, thank you for the help.
I cloned the current repository and ran the following command:
python demo.py --prompt "Photo of Emilia Clarke with a bright red hair" --init-image ./data/input.png --strength 0.5
Full traceback:
Traceback (most recent call last): File "demo.py", line 74, in <module> main(args) File "demo.py", line 43, in main eta = args.eta File "/cnvrg/test/stable_diffusion.openvino/stable_diffusion_engine.py", line 155, in __call__ latents = self.scheduler.add_noise(init_latents, noise, timesteps)[0] AttributeError: 'PNDMScheduler' object has no attribute 'add_noise'