Programmers have the consensus that linux and mac are better than windows. But I disagree. Everything works like a charm on windows, but in WSL it's all a nightmare.
For those unable to import TensorFlow into their notebook: Run "python3 -m pip install notebook" to install Jupyter in the same environment as TensorFlow. The sudo apt version of Jupyter may not use the same Python environment. This worked for me.
Hey guys if you get an empty list after this commmand :python3 -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))" You can try to downgrade your tensorflow version to 2.15 by running this command: pip install tensorflow==2.15 it worked for me
not working i think the problem is where the anaconda is getting installed there python is not got installed and when i try to run my code anywhere it says "tensorflow is not added"
It feels like it works, the computer doesn't use 100% of its capacity to computer anymore. But the GPU usage still seem to be very low (8%). Maybe it's the training task that doesn't allow for enough parallelization, I'll keep trying.
My best guess is that all the errors at 10:50, which are the same I have, indicated the computer is still NOT able to use GPU and can only use CPU, which defeats the whole purpose of going into wsl in the first place as the standalone Windows does the same...
Not worked for me: tensorflow/core/common_runtime/gpu/gpu_device.cc:2251] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU
For those having the error " tensorflow/core/common_runtime/gpu/gpu_device.cc:2251] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU " when listing the GPUs, I fixed it by downgrading tensorflow to 2.15 version by running the command pip install tensorflow==2.15 ; then I saw another problem : it appeared that my GPU was not detected by the listing, to fix this, just go to your nvidia panel -> 3D parameters -> manage 3D parameters and select your GPU in the preferred Graphic Processor Option.
He can find tensorflow when I run in terminal but when i move to jupyter he can not find. He say "No module named 'tensorflow'", and vedio don't run tensorflow in jupyter as example either. Can anyone help me please? maybe is this error : [IPKernelApp] ERROR | No such comm target registered: jupyter.widget.version I found out that the version of python in terminal is different with the version in jupyter?why?
I have an issue whereby all that you show works if I run ubuntu app, but not from cmd -> wsl, then conda is not discovered, but if I start ubuntu it is there
For some reason I keep getting Out of Memory error when I try with WSL. Tensorflow (GPU) works perfectly with Windows, however I get the out of memory error when I try Tensorflow (GPU) on WSL. appreciate your guidance
Can you help me? I tried to import tensorflow in jupyter notebbok, but mistaked appeared "No module named tensorflow". I tried to use pip install in notebook, it said that Requirement already satisfied. What is the problem?
Ok. Just to confirm. Activate your environment then type there python. Once you enter in python shell. Import tensorflow there. Let me know if that works.
worked almost perfectly, just had to run some pip install tensorflow --upgrafe to make it work. Is it possible to connect my VSC instance on my desktop to my laptop and use my GPU from remote?
it says can not locate jupyter~notebook package and couldnt install (py310) root@name:~# jupyter~notebook Command 'jupyter~notebook' not found, did you mean: command 'jupyter-notebook' from deb jupyter-notebook (6.4.8-1ubuntu0.1) Try: apt install (py310) root@name:~# sudo apt install jupyter~notebook Reading package lists... Done Building dependency tree... Done Reading state information... Done E: Unable to locate package jupyter~notebook
@@iagirre yeah bro actually gup 0 is nvidia by default, but in task manager it's shows GPU 0 is Intel and GPU 1 is nvidia this is different okay.. And if u get GPU 0 after running whole commonds in powershell then u can good to go, and if u want to run jupyter notebook or vs code then use this below command code . Then it will download vs code for u and after that u can execute ur .py and .ipynb file in it. After this click on run, then select tf-gpu as ur environment. That's it
Hi, Thanks for this wonderful video. I need some help. I followed the tutorial and everything worked except when you import the pandas ? I installed the same way like you did, but even after that it is showing "ModuleNotFoundError: No module named 'pandas'" Can you explain me whats the problem here ?
Any new package installed is not being found in jupyter-notebook. Try another one, sklearn for example. However, if you check on the terminal (with the rigth enviroment) it is there. I am openning the jupyter-notebook on the rigth enviroment as well. So, what is wrong? And, what is the solution?
i spent 7 days on trying to install this cancer, if this video doesnt help i will delete everything you did not show how you install cudnn, what versions of things, i am impressed this worked but i am deep in the toilet with the setup i give up
i figured it out thank you for offering, installing this stuff alone is way too much effort, good luck to anyone trying to work with this outside of stragight forward school projects@@KGPTalkie
Hello your video really helped me but ended kidna early. How to import the Kernel in VSC? And do i need an extra command to run my model with my GPU? And where do i find my folder cause i wanna import a csv.
Thank you! it worked for me without any errors😁. I followed through another video but it was 9 month old and got so many errors. But I have one problem. All libraries are importing fine but as I compile my model the kernel dies. Is it because of insufficient memory as I have 8 gigs ram and by default wsl is allocated 4 gigs.
@@Songoku-hg6bc Yes, in my case RAM was the problem. To make sure you can check error messages or check the computer resources while running your model. You can check if your model works on Google Colab, it offers good understandability of computing resources.
fuck tensorflow it's not working after trying for 10 fuckin hours
Bro, took me several days and still nothing
Brother US
did you find any solution , I have been trying for days
me working from 3 days bro. fuck it
Programmers have the consensus that linux and mac are better than windows. But I disagree. Everything works like a charm on windows, but in WSL it's all a nightmare.
For those unable to import TensorFlow into their notebook: Run "python3 -m pip install notebook" to install Jupyter in the same environment as TensorFlow. The sudo apt version of Jupyter may not use the same Python environment. This worked for me.
Good catch!
Hey guys if you get an empty list after this commmand :python3 -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
You can try to downgrade your tensorflow version to 2.15 by running this command: pip install tensorflow==2.15 it worked for me
Thanks a lot brother It worked for me
Thanks a lot for this comment. It saved me
YOU MADE MY DAY!!
Thanks buddy! It worked for me!! Such a relief😊😊
thank youuuuu😘
thank you!! you have no idea how long i've struggled with this before i found this video :)
Thanks a lot! This is the only video that correctly and efficiently explains the process.
THANK YOU! This finally got me running TensorFlow on windows
Helped me a lot man, more than +3 hours trying to fix it by myself and you came like god from heaven to teach me how to. Thank you!!!
Bro this is the best explanation of how to install tensorflow GPU in WSL.... 🌟🌟🌟🌟
hello after the import tensorflow step its telling me to that TENSOR RT Not found , how do i fix this
I am have same problem
It is optional to install this library. If you want to install it check out the documentation
Thanks for such an accurate information. It'll just take time to setup wsl2. Remaining all steps are very easy.
not working
i think the problem is where the anaconda is getting installed there python is not got installed and when i try to run my code anywhere it says "tensorflow is not added"
Did you follow all steps? As described in the video description.
Very nice vid bro it helped me a lot.
OMG thank you so much saved a lot of time :)
Thanks
It feels like it works, the computer doesn't use 100% of its capacity to computer anymore. But the GPU usage still seem to be very low (8%). Maybe it's the training task that doesn't allow for enough parallelization, I'll keep trying.
My best guess is that all the errors at 10:50, which are the same I have, indicated the computer is still NOT able to use GPU and can only use CPU, which defeats the whole purpose of going into wsl in the first place as the standalone Windows does the same...
Not worked for me:
tensorflow/core/common_runtime/gpu/gpu_device.cc:2251] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU
same here
Same here, did anyone managed to fix this ?
May I know where you installed CUDA toolkit and cuDNN, inside or outside WSL2?
Inside wsl2.
For those having the error " tensorflow/core/common_runtime/gpu/gpu_device.cc:2251] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU " when listing the GPUs, I fixed it by downgrading tensorflow to 2.15 version by running the command pip install tensorflow==2.15 ; then I saw another problem : it appeared that my GPU was not detected by the listing, to fix this, just go to your nvidia panel -> 3D parameters -> manage 3D parameters and select your GPU in the preferred Graphic Processor Option.
Thanks bro it worked for me
When i try to import tensorflow module in notebook it shows error- "no module named tensorflow"
Did you fix it?
Thx very much, working 100%
He can find tensorflow when I run in terminal but when i move to jupyter he can not find. He say "No module named 'tensorflow'", and vedio don't run tensorflow in jupyter as example either. Can anyone help me please?
maybe is this error : [IPKernelApp] ERROR | No such comm target registered: jupyter.widget.version
I found out that the version of python in terminal is different with the version in jupyter?why?
"On WSL" is not shown when setting the interpreter in Pycharm. I am stucked. Any idea? Thanks
can we work on pycharm by this ? as you are downloading jupyter notebook can i write pycharm?
incredible! thank you!
I have an issue whereby all that you show works if I run ubuntu app, but not from cmd -> wsl, then conda is not discovered, but if I start ubuntu it is there
At 11 minutes and 20 seconds in the video, the prompt: Unable
to register cudnn... that is it normal?
Brdr I followed all the steps and everything looked the same as in the video but still when i tried to run gpu in my jupyter notebook it tells no GPU
I am using windows native also my gpu discoverable but it shows jit compilation failed when running some code
For some reason I keep getting Out of Memory error when I try with WSL. Tensorflow (GPU) works perfectly with Windows, however I get the out of memory error when I try Tensorflow (GPU) on WSL. appreciate your guidance
I am also facing same kind of problem. If you could solve it please educate me
Can you help me? I tried to import tensorflow in jupyter notebbok, but mistaked appeared "No module named tensorflow". I tried to use pip install in notebook, it said that Requirement already satisfied. What is the problem?
Did you find solution ?
Which environment are you using? Make sure you have activated your environment before using jupyter notebook. Use conda activate base
@@KGPTalkie yes i followed your every step also "conda activate .....". Still I'm facing this issue
Ok. Just to confirm. Activate your environment then type there python.
Once you enter in python shell. Import tensorflow there. Let me know if that works.
@@KGPTalkie I did as you said in powershell and tensorflow is running but when I run it in jupyter-notebook it is showing "No module named".
Thanks. I wanna buy you a coffee ☕
awesome video
Cannot we install tensorflow directly in our windows environment?
from tf 2.11 not anymore
please help mine returns an empty list and i have tried all the possible solutions i saw online I need help please
Thank you so much man!
worked almost perfectly, just had to run some pip install tensorflow --upgrafe to make it work. Is it possible to connect my VSC instance on my desktop to my laptop and use my GPU from remote?
Use ssh in terminal within vs code from remote laptop to connect to remote machine
it says can not locate jupyter~notebook package and couldnt install
(py310) root@name:~# jupyter~notebook
Command 'jupyter~notebook' not found, did you mean:
command 'jupyter-notebook' from deb jupyter-notebook (6.4.8-1ubuntu0.1)
Try: apt install
(py310) root@name:~# sudo apt install jupyter~notebook
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
E: Unable to locate package jupyter~notebook
What about nvidia driver
Is it same in other case of tensorflow or i have to install other os based driver
Windows version. you need to install on your windows pc.
It's showing gpu 0 which is my intel uhd graphic card but i want to use my gpu 1 which is nvidia graphic card,,...
How to switch to gpu 1
I have the same issue, did you solve it?
@@iagirre yeah bro actually gup 0 is nvidia by default, but in task manager it's shows GPU 0 is Intel and GPU 1 is nvidia this is different okay..
And if u get GPU 0 after running whole commonds in powershell then u can good to go, and if u want to run jupyter notebook or vs code then use this below command
code .
Then it will download vs code for u and after that u can execute ur .py and .ipynb file in it.
After this click on run, then select tf-gpu as ur environment. That's it
@@EshwarShiva-v3x in my case I solved this by choosing the nvidia gpu as the default one in nvidia control panel
Hi, Thanks for this wonderful video. I need some help. I followed the tutorial and everything worked except when you import the pandas ? I installed the same way like you did, but even after that it is showing "ModuleNotFoundError: No module named 'pandas'" Can you explain me whats the problem here ?
Select the environment in which you installed pandas and then try to import pandas
Any new package installed is not being found in jupyter-notebook. Try another one, sklearn for example. However, if you check on the terminal (with the rigth enviroment) it is there. I am openning the jupyter-notebook on the rigth enviroment as well. So, what is wrong? And, what is the solution?
no matches found: tensorflow[and-cuda]
i spent 7 days on trying to install this cancer, if this video doesnt help i will delete everything
you did not show how you install cudnn, what versions of things, i am impressed this worked but i am deep in the toilet with the setup i give up
Don't worry. If this doesn't help let me know. I can connect and help you
i figured it out thank you for offering, installing this stuff alone is way too much effort, good luck to anyone trying to work with this outside of stragight forward school projects@@KGPTalkie
Hello your video really helped me but ended kidna early.
How to import the Kernel in VSC? And do i need an extra command to run my model with my GPU?
And where do i find my folder cause i wanna import a csv.
Please watch last 2 minutes carefully. I have shown this.
Brother nano .bashrc
I can't get same output
Please share timestamp of the video where you are matching content. And what is your file's content
7:59
sudo command is not working can you help me with that
What error are you getting? You can proceed without sudo as well.
thank you very much . How can I connect it to Spyder?
Thank you! it worked for me without any errors😁. I followed through another video but it was 9 month old and got so many errors.
But I have one problem. All libraries are importing fine but as I compile my model the kernel dies. Is it because of insufficient memory as I have 8 gigs ram and by default wsl is allocated 4 gigs.
did you fix your error? I am facing the same error, can you tell me if you fixed it
@@Songoku-hg6bc Yes, in my case RAM was the problem. To make sure you can check error messages or check the computer resources while running your model.
You can check if your model works on Google Colab, it offers good understandability of computing resources.
@@Songoku-hg6bc I am facing the same error. The kernel dies
Its Show GPU : 0 which means it not Taking ke GPU1 MAIN GPU
How to use GPU1 ?
Same question.....
Simply install , Tensorflow
This Video is fine but it didn't detect Main GPU
@@EshwarShiva-v3x
If someone has problems with internet. Don't use USB internet from phone. Use phone internet with Wifi i.e hotspot!
Thanks for sharing this
Nice tutorial. I have veen looking for a way to use WSL with VS Code. Thanks.
it did not work
This tutorial doesn't even work. Kinda garbage.
dookie ahh tutorial
errors kon tera papa yani mai solve karunga
Please mention which error and at what timestamp? Won't disturb my father for this small issue. Btw thanks for watching.
Brdr I followed all the steps and everything looked the same as in the video but still when i tried to run gpu in my jupyter notebook it tells no GPU
Thanks a lot man!!