joaolages / diffusers-interpret Goto Github PK

Notebook loading error
There was an error loading this notebook. Ensure that the file is accessible and try again.

Invalid Credentials

Problably the notebook it is not set to public.

I received this error, cannot import name '_png' from 'matplotlib' when trying to run the new plot command when using Google Colab.

It seems Colab has an older version of matplotlib running - 3.2.2
If I forced an install of matplotlib, pip install -U matplotlib and restarted the runtime, I could get the plot to work correctly.

Plot is helpful!

Hi, I added it as a comment in previous issue, but I think it's nice to raise it as a separate ticket - an issue there was it'd be nice to update diffusers package to newer version, but nowadays I am unable even to run colab demo due to this error.

I wanted to follow up as presented demo colab is not working at all. When downloading StableDiffusion pipeline I got error

TypeError: getattr(): attribute name must be string

I found an error about it on SO: https://stackoverflow.com/questions/74687769/typeerror-getattr-attribute-name-must-be-string-in-pytorch-diffusers-how

When I force to use 0.24, I got error on importing diffusers_interpret:
ImportError: cannot import name 'preprocess_mask' from 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_inpaint' (/usr/local/lib/python3.10/dist-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py)

So I tried to install it with pip again but it downloads diffusers back to 0.3.0

So I can create a pipe for stable diffusion (requires diffusers > 0.4) OR I can use diffusers_interpret (requires 0.3).
Both of them does not work together in demo colab meaning I cannot reproduce your output

When using diffusers-interpret with the latest diffusers (0.8.0, yes, I need this version because I use Euler discrete scheduler), it gives the following error:

ImportError: cannot import name 'preprocess_mask' from 'diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_inpaint' (/usr/local/lib/python3.7/dist-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py)

Can it be fixed to work with it? 0.3.0 is now very outdated.
Thanks!

Version 0.3.0 of the 🤗 Difusers introduces enable_attention_slicing, and I wonder if there's a way to implement this in the explainer. Below is the code that I used and it ran out of CUDA memory:

# Import pipeline
import torch
from diffusers import StableDiffusionPipeline

torch_device = "cuda" if torch.cuda.is_available() else "cpu"

pipe = StableDiffusionPipeline.from_pretrained(
    "CompVis/stable-diffusion-v1-4",
    use_auth_token = True,
    revision = "fp16" if torch_device != "cpu" else None,
    torch_dtype = torch.float16 if torch_device != "cpu" else None)

pipe.to(torch_device)

pipe.enable_attention_slicing() # attention optimization for less memory usage

# Pass pipeline to the explainer class

from diffusers_interpret import StableDiffusionPipelineExplainer

explainer = StableDiffusionPipelineExplainer(pipe)

prompt = "photograph, piggy, corn salad"

with torch.autocast(torch_device):
    output = explainer(prompt,
                       guidance_scale=7.5,
                       num_inference_steps=17)

output.image

When I try the following code snippet from your notebook I get a TypeError

prompt = "A cute corgi with the Eiffel Tower in the background"

generator = torch.Generator(device).manual_seed(2022)
with torch.autocast('cuda') if device == 'cuda' else nullcontext():
    output = explainer(
        prompt, 
        num_inference_steps=15, 
        generator=generator
    )

TypeError: '>' not supported between instances of 'NoneType' and 'int'

What would you like your documentation website to look like? I would like to help creating and/or maintaining it.
I have some experience with Github Pages built via Jekyll. If you want something similar, would you mind creating a gh-pages branch and choose a Jekyll theme of choice?