jiauzhang / gigagan Goto Github PK
View Code? Open in Web Editor NEWImplementation of GigaGAN: Scaling up GANs for Text-to-Image Synthesis
License: MIT License
Implementation of GigaGAN: Scaling up GANs for Text-to-Image Synthesis
License: MIT License
The error is IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
at images = g(z, text_embeds)[0]
statement.
rahulbhalley@192 GigaGAN % python3 generate.py
Traceback (most recent call last):
File "/Users/rahulbhalley/Desktop/Hide-and-Seek/image-synth/GigaGAN/generate.py", line 17, in <module>
images = g(z, text_embeds)[0]
^^^^^^^^^^^^^^^^^
File "/opt/homebrew/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/rahulbhalley/Desktop/Hide-and-Seek/image-synth/GigaGAN/model.py", line 140, in forward
styles = [self.style(s) for s in styles]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/rahulbhalley/Desktop/Hide-and-Seek/image-synth/GigaGAN/model.py", line 140, in <listcomp>
styles = [self.style(s) for s in styles]
^^^^^^^^^^^^^
File "/opt/homebrew/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/homebrew/lib/python3.11/site-packages/torch/nn/modules/container.py", line 217, in forward
input = module(input)
^^^^^^^^^^^^^
File "/opt/homebrew/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/Users/rahulbhalley/Desktop/Hide-and-Seek/image-synth/GigaGAN/layers.py", line 11, in forward
return input * torch.rsqrt(torch.mean(input ** 2, dim=1, keepdim=True) + 1e-8)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
Hello @JiauZhang - I was on StyleClips repo and we were trying to confirm something.
I need to train a new SG model, with our own CLIP "captions" and not use LAION or anything else.
So, my alpha test case was to build a new SG model with images of Men and Women.
My question is - Did you train a new model with your own "captions" or how are you adding NEW keywords/captions?
In our case we have our own internal nomenclature which we must train into some medical images for a use-case, and we have to figure out how to do this.
Hello, great effort on this re-implementation.
Would you need GPU resources / other assistance to complete your work? How far are you from achieving the paper results for the upscaler?
Hello, I only saw the definition of clip loss in clip.py. I didn't find where you used it? But your paper points out that this loss is used.
When use_text_cond is True , the text encoder works but text embeds cannot be concatenate with image features in model.py styles = [torch.cat([style_, t_global], dim=1) for style_ in styles]
, these tensors are not in same size, how should I modify this problem, thanks.
I found adding the cross attention module would make the training become unstable. The G loss raised to NaN.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.