gnobitab / fusedream Goto Github PK

"😵 Uh oh! This model can't be run on Replicate because it was built with a version of Cog that is no longer supported"
https://replicate.com/gnobitab/fusedream

Hi,

When I run the cell of the Colab notebook that contains the lines:

!gdown 'https://drive.google.com/uc?id=17ymX6rhsgHDZw_g5XgAFW4xLSDocARCM'
!gdown 'https://drive.google.com/uc?id=1sOZ9og9kJLsqMNhaDnPJgzVsBZQ1sjZ5'

I get the following error:

Excuse me,

I tried to reproduce the CLIP R-precision in the paper by using the code in DM-GAN.

However, I got 51 for AttnGAN, 65 for DM-GAN and 37 for DF-GAN with the official model weight.

I would appreciate if you could provide the code to compute CLIP R-precision.

Hello, I couldn't find an option for saving the image before completion (every N iterations) like in the original Big Sleep.

I find this option very useful because this allows me to stream the image updates and by interpolating the frames I can have a live and smooth view of the current result as it's being created.

So could you add this option? Thanks!

Hi! Great paper :) So the citation [20] should actually go to Ryan Murdock (https://twitter.com/advadnoun), who discovered how to put CLIP and BigGAN together in a colab notebook. I simply packaged it into a CLI tool.

Some results
python fusedream_generator.py --text 'A photo of a blue dog.' --seed 1234

python fusedream_generator.py --text 'A boy is reading papers.' --seed 1234

Hello, in the original Big Sleep this option is very helpful as it helps in tweaking the quality of the result.

Could this be added to FuseDream?

Or at least could you please point me to the correct line to change, if I wanted to modify the default setting?

Thanks!

Hi,

Thanks for the great work and I'm impressed with your work.
I want to know when the FuseDream-Composition will be released.

Thank you.

how generate different image for same text

Hi, when I run command 'python fusedream_generator.py --text 'A photo of a blue dog.' --seed 1234' , there comes up an error. Why the clip library doesn't have "load" attribute? Does someone have same issue? My clip version is 0.2.0

Generating: A photo of a blue dog.
[8, 16, 32, 64, 128, 256, 512] {8: False, 16: False, 32: False, 64: True, 128: False, 256: False, 512: False}
Adding attention layer in G at resolution 64
Number of parameters: 82296804
Traceback (most recent call last):
File "fusedream_generator.py", line 25, in
generator = FuseDreamBaseGenerator(G, config, 10)
File "/home/ubuntu/supy/FuseDream-main/fusedream_utils.py", line 150, in init
self.clip_model, _ = clip.load(clip_mode, device=device)
AttributeError: module 'clip' has no attribute 'load'

List of non-documented dependencies found:

1. h5py
2. clip

Hello, author, do this code support generating batch images of batch text at the same time? Instead of just generating one image for one text

Hello, in the original Big Sleep this option is very helpful as it helps in tweaking the quality of the result.

Could this be added to FuseDream?

Or at least could you please point me to the correct line to change, if I wanted to modify the default setting?

Thanks!

Hi,

Thanks for the great work. Is the implementation of FuseDream-Composition going to be available as well?
I don't see the composition described in the paper anywhere in the code.