-
Notifications
You must be signed in to change notification settings - Fork 10
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
struggling when trying to install waifu2x-caffe on google colab #9
Comments
This error occurs with older versions of CMake (FindCUDA) that do not support CUDA 10 or later. You should upgrade CMake. |
thank you @nagadomi, that old CMake version was the problem, I installed Cmake with these commands:
then I deleted /usr/local/bin/cmake, and ran the following command:
outputs:
I then ran
however, cuDNN was not found, although I have installed it using these commands:
and
the current Cuda version is 11.1 so I downloaded cuDNN for for Cuda 11.1. this is the output after running
does it look right to you? |
after opening video2x.yaml and changing from
to this
I tried to run video2x with waifu2x_caffe:
but I got this:
how to get it to work? |
Set
Python's |
Thank you @nagadomi for your reply, could you tell me where I should place If I run these two commands in this directory
I get this output
and If I do this:
and get the same output as before, which is still:
with |
thank you @nagadomi again, I updated
it gives a weird output:
if I try to run waifu2x-caffe through video2x, I get this:
looks like there's a problem with waifu2x-caffe |
It seems that cudnn is detected, but for some reason it fails to get the version.
waifu2x-caffe uses CP932 encoding, not utf-8.
Also, the error is probably due to the fact that the |
from https://forums.developer.nvidia.com/t/cudnn-found-but-version-cant-be-deduced/145551/5
You need to build with CUDA10 and cuDNN7 or fix it yourself. |
thank you @nagadomi for your fast reply!, i opened and edited
into:
I then created a
because I found that
(lltcggie-caffe is I then tried to
and running
although cuDNN was found:
there were lots of warnings Edit:
|
It's a problem that is incompatible with cuDNN8. lltcggie/caffe@ed6cff4 |
@nagadomi, I may have used the wrong caffe, which one should I use and place under |
for the whole time I downloaded caffe from https://github.com/nagadomi/caffe, and placed inside Edit:
and If I use that
Edit2:
I suppose I have to restart notebook and redo everything. |
https://github.com/lltcggie/caffe is the original lltcggie/caffe, which is intended to be built on Windows Visual Studio. |
@nagadomi , with your patched
this is from
and this is from
|
I have made changes to support CUDA11, cuDNN8, and OpenCV4.
|
thanks for your reply @nagadomi , this is the output of
with a lot of |
and this is the output of
|
I assume you have copied cudnn.h and libcudnn.so to /usr/local/cuda/, but that is not necessary.
It seems that it has finally been found.
The build finished without error. I think it will work. |
Thank you so much @nagadomi , it finally worked, it successfully upscaled my first picture, using cudnn. Although there are warnings after running |
Hi!, I'm not sure if I can ask this here but, I'm pretty new to Github stuffs, can anyone teach me install waifu2x-caffe on Google Colab from scratch? Thank you so much! Waifu2x is an awesome project, I really appreciate everything you guys have done for this project and anything related. It's just that my laptop is too weak to run waifu2x, so I'm trying to run it on Google Colab. hope anyone would help me complete this notebook and I'll definitely share it once it's finished and working!
Edit:
I've read through the issue at #3 posted by @sjscotti, this is what I've done so far:
running the above block of codes gave me this:
what should I do to make waifu-caffe work?
The text was updated successfully, but these errors were encountered: