Tenserflow GPU (Latest 2.14) installation on Windows 11 through WSL2

[@anooshkabajaj]

This solved my problem! From the past few days I went through a lot of tutorials but still wasn't able to setup. Your tutorial worked!!

[@vohoang6693]

Thank you my god, you save me . I have install and run with tensorflow 2.16.1. The section from 11:34 to 14:58 is outdate.If tensorflow version you want to install, requirement cudnn version 8, you can run sudo apt install libcudnn8 && \ sudo apt install libcudnn8-dev && \ sudo apt install libcudnn8-samples instead

[@neiderwong]

thanks, it worked :)

[@igoralbuquerque7251]

You install CUDA 12.3 or 12.2?

[@vohoang6693]

@@igoralbuquerque7251 I install CUDA 12.2 and it's still work

[@vohoang6693]

@@igoralbuquerque7251 12.3

[@CASPER.410]

Thank you very much for this video. I had trouble setting up the tensorflow in my computer and I couldn't get it to work for 3 weeks and this video finally helped me to install this successfully! I really want to appreciate this video!

[@learn_with_zoro]

That's fantastic to hear! I'm thrilled the video helped you the TensorFlow setup challenge. If you found it valuable, consider subscribing for more tech tips and tutorials. Happy coding!

[@CASPER.410]

@TechTestingN of course I did subscribe it and liked the video. I'll wait for more tips and tutorials! Thanks again

[@learn_with_zoro]

Thank you

[@vishwaskotegar5248]

beautiful video. just a small suggestion since i ran into a problem and found a workaround . the issue was at step 5.6 sudo apt-get install libcudnn8=8.x.x.x-1+cudaX.Y if you face the same problem check the available packages - apt list -a libcudnn8 and then rename accordingly. i had installed cuda12.3 but this package was unavailalble. and i installed cuDNN for 12.2 while still primarily using 12.3 .. somehow works for me

[@HyagoBragaDosSantos]

Thank you so much for this video. I was "lost" into a deep research about solutions for my problem in which I didn't get to use my GPU for CNN training. Thank you again!

[@CHIH3B]

i installed the newer tensorflow version with this and it worked.Thank you

[@learn_with_zoro]

You’re welcome. I’m glad to hear that it worked for you.

[@GuilhermeVieiraSechat]

Thansk! This worked perfectly for me, after 11h trying. I beliece the libraries were the issue of my tries before (starting directly from tensorflow[and-cuda]

[@euuIgor]

You are amazing. I've tried multiple tutorials and none of them worked. Your tutorial was the only one that worked, I'm so happy, thank you so much!!!😁

[@juanmanuelcorino1717]

You are de Best. The only change is install tensorflow 2.15

[@mehmetsarkaya5267]

Thanks man! This video helped me a lot!!!!

[@SolomonChika]

Thank you so much, I have spent days trying to get this to work. I am really grateful.

[@Jim99O]

Great video saved me a lot of time and nerves :)

[@learn_with_zoro]

Great to hear!

[@ramanr8928]

Excellent , thank you for this awesome guide! Subscribed to see more good videos from you, cheers!

[@learn_with_zoro]

Awesome, thank you!

[@giuseppetricarico8724]

When I try to issue "python3 -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"" i receive the following error "ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject". I checked all step and seems I did evwerything as in the video..

[@KnowledgeIsGold]

Hi, i followed your tutorial. But i can't import from tensorflow.keras.preprocessing.image import ImageDataGenerator. It says module not fond

[@kuleenthejas8382]

when i try step 7 the tensorflow version is 2.17 but later when i try step 8 the tensorflow version finally ends up being 2.17 instead of 2.14 this doesn't seem to change i am getting error in the step 9 pls suggest something regarding this

[@AdeelZia-k2t]

I don't understand after doing everything why did you do tensorflow -cuda at the end, It ended up upating my cuda to 12.3 while everything else was of 11.8 all along and now I have to do it all over again.

[@augustojlle]

Thank you bro

[@learn_with_zoro]

Good luck!

[@dinos0h]

This works, omg. thank you so much

[@ramaadillazein9786]

thanks, your video is helpful

[@learn_with_zoro]

Thankyou 😊

[@guilhermesantos3661]

Works perfect!!!

[@clp-d7r]

when i try to import CUDA GPG Key in 11:08 this message pops out "cp: cannot stat '/var/cudnn-local-repo-*/cudnn-local-*-keyring.gpg': No such file or directory" how can i fix this?

[@fuzesky4401]

How can I use a Jupyter notebook because when I create this file I get access denied?

[@400bug4]

excellent tutorial👍

[@learn_with_zoro]

Glad you liked it!❤️

[@arandomguy3526]

Is installind cuda toolkit and cudnn necessary? As I only did pip install tensorflow[-andcuda] and it is working fine. Why did u show installation of cuda toolkit and cudnn if it was not necessary amd could only be done using tensorflow[-andcuda] ?

[@AdeelZia-k2t]

For older hardware, we have to lock in versions because directly it pulls the latest versions which might not be compatible

[@кириллполонский-е6ф]

man you are the best

[@dealwithpresent7988]

is this same work for wsl Kali linux

[@ramaadillazein9786]

The cuda toolkit is installed in the py31 env right? So if I want to install cudf which require new environment, I have to install cuda toolkit on my new environment (for cudf), right?

[@learn_with_zoro]

I think yes, I am not sure either.

[@h4wk_0]

guide is apparently nice. i have a question. why is my gpu slower than cpu when i am starting to train my neural network. i have 1660 ti and i7-10750h. feel awful cuz i waste 2 hourse to install that shit

[@learn_with_zoro]

It depends on exactly what your training. and compatibility of that algorithm with GPU. let me explain somethings GPUs are more efficient than CPUs for parallel computations, which are common in neural networks. However, some types of neural networks, such as recurrent neural networks (RNNs) or long short-term memory (LSTM) networks, are more sequential in nature and cannot be easily parallelized. This can limit the speedup of using a GPU over a CPU and There is a cost associated with transferring data between the CPU and the GPU, which can be significant if the data size is large or the frequency is high. This can reduce the performance gain of using a GPU over a CPU. I hope you understand it.

[@h4wk_0]

@@learn_with_zoro yeah, thx. I’ve already got it

[@minhnguyen1338]

thank you very much

[@learn_with_zoro]

You are welcome

[@vegam05]

Thankyou!

[@AniketKumarSinha-xd6xu]

Thnx broooooooooooo 😭😭😭😭

[@ChristianRichardson-i5f]

fun fact: check if wsl is installed already before installing it, or you might brick your computer

[@learn_with_zoro]

I think it's not in all the cases. But still Check if WSL is already installed before installing it, or you may corrupt Windows and brick your computer. To check, open PowerShell as admin and type wsl --list --verbose. If you see Linux distributions, WSL is installed. If you see an error, WSL is not installed. To install WSL, follow the official instructions.

[@joaquincampo6829]

I love you

[@d.156]

Thank you so much!! this finally worked 🥲