maua-maua-maua / maua Goto Github PK
View Code? Open in Web Editor NEWDeep learning toolkit for image, video, and audio synthesis
License: GNU General Public License v3.0
Deep learning toolkit for image, video, and audio synthesis
License: GNU General Public License v3.0
I'm trying to upscale some images and am trying to use this repo to do so.
I have run into a couple of issues.
During the installation steps:
git clone --recursive https://github.com/maua-maua-maua/maua.git
cd maua
conda create -n maua python=3.8 pytorch torchvision torchaudio cudatoolkit=11.3 cudatoolkit-dev=11.3 cudnn mpi4py Cython pip=21.3.1 -c nvidia -c pytorch -c conda-forge
conda activate maua
pip install -r requirements.txt
pip install -r audio/requirements.txt
pip install cupy-cuda113==9.6
When running pip install -r requirements.txt
, the process gets stuck when attempting to build the wheel for nestedtensor
.
I commented this out hoping that for upscaling, this package will not be required.
When I tried to run the provided upscaling command,
python -m maua super /path_to_my_image.png --model_name RealESRGAN-pbaylies-hr-paintings
I got an error message because pytorch-lightning
wasn't installed.
I resolved this by installing the package from here:
https://www.pytorchlightning.ai/
I then reran the upscaling command but got the following error:
Segmentation fault (core dumped)
I am using Ubuntu 18.04. I have a Tesla T4 gpu. My CUDA version is 11.4 and my NVIDIA driver version is 470.82.00.
When this error occurred, my cpu was about 98% idle, I had about 10Gb of RAM free, and my T4 was completely free.
Could you please advise?
when processing lengthy (at least few minutes) videos with diffusion, the process gets slowing down progressively (probably taking days instead of hours). it may be due to the big optflow data files (in my case ~5gb); may it be more optimal to save them the per frame?
The new fast+efficient .mflo encoding of optical flow leads to vertical and horizontal line artifacts due to the magnitude channel interfering with JPEG encoding
Keep seeing this repeated error on install
Running command git submodule update --init --recursive -q
error: unable to create file docs/d0/d14/structnlohmann_1_1detail_1_1is__compatible__integer__type__impl_3_01RealIntegerType_00_01Compatie5920c849e839ebb9f8c57349c900796.html: Filename too long
_attn/cutlass'
error: subprocess-exited-with-error
× git submodule update --init --recursive -q did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
When I try to do the basic install I run into a repeated error:
Running command git submodule update --init --recursive -q
error: unable to create file docs/d0/d14/structnlohmann_1_1detail_1_1is__compatible__integer__type__impl_3_01RealIntegerType_00_01Compatie5920c849e839ebb9f8c57349c900796.html: Filename too long
_attn/cutlass'
error: subprocess-exited-with-error
× git submodule update --init --recursive -q did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
Hi! I'm trying to get this project running specifically for the audio-visual part. I'm following the installation instructions in the root README, running in a Google Colab notebook using a T4 GPU.
I've run the following:
!pip install nvidia-cudnn-cu116 --extra-index-url https://pypi.ngc.nvidia.com
as per the earlier issue here.
Then:
!pip install numpy Cython torch --extra-index-url https://download.pytorch.org/whl/cu116
and finally:
!pip install git+https://github.com/maua-maua-maua/maua.git --extra-index-url https://pypi.ngc.nvidia.com --extra-index-url https://download.pytorch.org/whl/cu116
Which then gives the output:
Looking in indexes: https://pypi.org/simple, https://pypi.ngc.nvidia.com/, https://download.pytorch.org/whl/cu116
Collecting git+https://github.com/maua-maua-maua/maua.git
Cloning https://github.com/maua-maua-maua/maua.git to /tmp/pip-req-build-0gqqay2x
Running command git clone --filter=blob:none --quiet https://github.com/maua-maua-maua/maua.git /tmp/pip-req-build-0gqqay2x
Resolved https://github.com/maua-maua-maua/maua.git to commit d968fd91f0d9b371ae5a5aa8e7ad9e9a906ded22
Running command git submodule update --init --recursive -q
Preparing metadata (setup.py) ... done
Collecting apex@ git+https://github.com/NVIDIA/apex (from maua==0.4)
Cloning https://github.com/NVIDIA/apex to /tmp/pip-install-hyxcyr_i/apex_ee92ddee1c1c471a964997c89113d2cc
Running command git clone --filter=blob:none --quiet https://github.com/NVIDIA/apex /tmp/pip-install-hyxcyr_i/apex_ee92ddee1c1c471a964997c89113d2cc
Resolved https://github.com/NVIDIA/apex to commit 141bbf1cf362d4ca4d94f4284393e91dda5105a5
Running command git submodule update --init --recursive -q
Installing build dependencies ... done
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
Getting requirements to build wheel ... error
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
The end game here is that I just want to be able to run the audio-visual code with a pretrained StyleGAN3 model preferably in Google Colab, so any solution which gets me there I'm happy with 😄 Here's a test environment for you.
is there a way to change the w/h of a video output? currently using an init video that is 1024 x 512, and when i try to change output size in the code, (512,512) it gives me this:
RuntimeError: The size of tensor a (512) must match the size of tensor b (1024) at non-singleton dimension 2
If i try changing either width or height I am presented with this. I may just be trying to modify the wrong param - apologies in advance!
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.