► My Other Tutorials: Build and Install OpenCV 4.5.1 With CUDA GPU Support on Windows 10 th-cam.com/video/YsmhKar8oOc/w-d-xo.html Face Detection Using OpenCV Python with CUDA GPU Acceleration th-cam.com/video/GXcy7Di1oys/w-d-xo.html YOLOv4 On Android Using TFLite th-cam.com/video/YzAjAS6Os8c/w-d-xo.html Custom YOLOv4 Object Detection with TensorFlow and TFLite th-cam.com/video/vzTCJM18uoM/w-d-xo.html Darknet YOLOv4 Custom Object Detection: Part 2 (Training YOLOv4 Darknet) th-cam.com/video/-NEB5P-SLi0/w-d-xo.html Darknet YOLOv4 Custom Object Detection: Part 1 (Preparing Custom Dataset) th-cam.com/video/sKDysNtnhJ4/w-d-xo.html YOLOv4 Object Detection with TensorFlow, TFLite and TensorRT th-cam.com/video/tCmC7nyfJp8/w-d-xo.html Darknet YOLOv4 Object Detection for Windows 10 on Images, Videos, and Webcams th-cam.com/video/FE2GBeKuqpc/w-d-xo.html Real Time Object Detection on Webcam and Videos Using OpenCV With YOLOv3 and YOLOv4 | Windows Linux th-cam.com/video/FjyF03uawsA/w-d-xo.html Build and Install OpenCV 4.4.0 with CUDA (GPU) Support on Windows 10 th-cam.com/video/tjXkW0-4gME/w-d-xo.html Install TensorFlow GPU and PyTorch with CUDA on Windows 10 Anaconda | CUDA 10.1 cuDNN 7.6 th-cam.com/video/PlW9zAg4cx8/w-d-xo.html Real-time Multiple Object Tracking with YOLOv4 TensorFlow and Deep Sort | Linux, Windows th-cam.com/video/GagII5PAeKg/w-d-xo.html
Second time i am going through this process. First time was a real hassle, this time knew where i should go directly. If you are watching this video to get the GPU running you are in the right place. Hands down the best tutorial ever.
After 2 days of a painful trial and error and rage process, I finally managed to get it working. It is a shame the official documentation is not as clear as you were in your explanation. Thanks, bro!
helped me! for me to get it to work I still had to install cuda (11.2) and cudnn (8.1.1.) manually but the pointer to the tf webpage on building on win from source definitely helped; all peachy now, thank you - gonna go watch the openCV tutorial
I cannot believe my eyes. After 4 hours of tearing my hair out trying to mess with conda environments / path variables / version control only for it to not work. I come across this video, a glimmer of hope, a mirage in the desert... To my surprise as I approach the oasis in the desert, it does not vanish. Alas, it is real, it amazes me how horrible the official documentation and pay-to-view blog-posts are about this process. You have saved me my sanity and for that I am grateful, will be sharing this with my friends. Now that my GPU is recognized now, Im getting an error: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize. Any tips?
This usually happens because of mismatch TensorFlow and cuDNN versions. Make sure you're using the same versions as mine. Sometimes, restarting the computer solves problem (if you have paths added to system variables, remove them).
Damn. Not sure why it only works this way nowadays but it worked. Why would I have to use pip on a conda environment? Why would I have to use conda-forge instead of regular conda distribution? Why wouldn't the official anaconda instructions work properly (as they did a few months ago???) Anyway, thanks dude. You saved my evening.
I am glad it worked for you. Regular conda distributions miss the latest versions sometimes. At the time of making this video, it was not possible to use them. Same is the case with pip. Only TF 2.4.0 was available with conda and v2.4.1 was available with pip. Official instructions often lag behind the changes that are made for every new version.
@@TheCodingBug Yea but at some point I had conda installed TF-GPU 2.3.0, CUDA 10.1 and cuDNN 7.6 and TF STILL wouldn't recognize my GPU. Tried different configurations, none worked even though it should on paper... Also following the official Anaconda guideliness for TF-GPU installation doesnt work anymore. Then I did the install your way and it works like a charm. I would expect Tensorflow and Conda to have way better documentation if it comes at least to the INSTALLATION of the library. :P
Good one 💪🏼 I’m getting error in import tensorflow as tf Traceback (most recent call last) File “”, line 1, in ModuleNotFoundError: No module named ‘tensorflow’ Any idea from your side
Sir, I got trouble with this, and I got this error: UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above. and this : 2021-03-05 21:08:13.831889: E tensor flow/stream_executor/cuda/cuda_dnn.cc:340] Error retrieving driver version: Unimplemented: kernel reported driver version not implemented on Windows and I don't know why i did exactly your procedure.
Thank you for the appreciation. I do not have an RTX card. But I am sure it will work. You can create a virtual environment and test it out. Do comment here if it works to help out others who might be wondering the same.
One question sir. I installed everything perfectly like you said in video & gpu is getting detected & working. However when I use gpu to train models, it only uses about 12% of gpu & improvement is not very significant over cpu. For comparison, I have R9 3900x & RTX 3080ti founder's edition. I ran a model once using CPU only & it takes 25s per epoch. If I use GPU, it takes 22s per epoch
@@TheCodingBug I did it. Still not much improvement. Can you share your GPU vs CPU result? I ran a gpu tensorflow google colab program (first link on google search) on my jupyter notebook & it showed that gpu is roughly 31 times less time taking cpu on a random 2D conv test, but when I ran it on my own model, the results were not that significant.
@@rayansh1245 it seems that for certain models (mostly ANNs) there won’t be any significant improvement if you use GPUs even though on testing with tensorflow, my GPU was on an average 32x faster than GPU. So I think this is how it’s supposed to work. But do tell me if there is a way.
Thanks for the video. I am able to install Tensorflow using Intel UHD GPU instead of dedicated NVIDIA GTX 1650 GPU and that to only when I open Anaconda prompt as administrator. How could I utilise my Nvidia graphics card in this scenario? Any help would be really appreciated. Thank you.
I didn't fully understand but if you have installed CUDA, cuDNN, latest Nvidia drivers, and TensorFlow GPU, it is going to pick nvidiw GPU, as shown at the end of this video. TensorFlow GPU does not utilize intel GPU. I assume you're working on a laptop. Make sure you are using performance mode and not battery saving mode.
@@TheCodingBug thanks for the reply. Yes i had installed the things as you mentioned above and it was similar to the end of the video but later when I tried the same using jupyter notebook it showed me device:0 as the name of GPU used using the command tf.test.gpu_device_name() and the same device:0 was listed as INTEL GPU in my task manager when i was looking for utilization.
After I try "import tensorflow as tf" to try it out, it tells me "ModuleNotFoundError: No module named 'tensorflow'" when I am sure I have followed all the steps properly. Please help me out :(
Very nice, but I get: Traceback (most recent call last): File "C:\Users atha\anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 64, in from tensorflow.python._pywrap_tensorflow_internal import * ImportError: DLL load failed while importing _pywrap_tensorflow_internal What am I doing wrong?
Its not showing my GPU. I followed your instruction to the letter. I installed the same version of Visual C++ Redist, the same version of tensorflow and the same cudnn and cudatoolkit. Its still not working.
Hello I am facing problem to using Premiere pro CC. Once I export my video and Unable to. Due to Error Compelling Move / GPU render error. Pls advice anything information needed to give me advice for solving the problem. Thanks in advance
I do the same as tutorial but i face this problem. Can you guide me to fix it.Thanks 2021-07-25 13:09:56.961151: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found 2021-07-25 13:09:56.961262: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
Thanks for your extremely helpful video. I don't have an appropriate graphics card on my home computer for tensorflow gpu, so I tried an Azure data science Windows virtual machine, which has a NVIDIA Tesla M60. Unfortunately, the tensorflow gpu version does not work on that Azure virtual machine. I upgraded to the latest Tesla M60 driver and I was able to follow your instructions to create a separate environment and install cudatoolkit, cudnn and tensorflow-gpu and get it to work.
hi, I'm trying to use tf 1.14 for this and I notice that cudnn=7.4 can't be installed from any channel, I already tried the manual way of installing cuda, cudnn with the correct version stated in the tf version table it just doesn't work, any idea? probably not but thanks anyway
i'm following your step and it works , gpu is detected by tensorflow . But when i start training the model i got the error with massage: UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above. [[node sequential_2/conv2d_6/Relu (defined at :1) ]] [Op:__inference_train_function_3151] Function call stack: train_function any solution?
@@gunnerstone120 ok i've already got the solution. There's a problem about memory management. U can insert this code : import tensorflow as tf config = tf.compat.v1.ConfigProto() config.gpu_options.allow_growth = True sess = tf.compat.v1.Session(config=config)
@@kikifade12 lol thanks for fast response, I was actually about to link you to this video that says same thing: th-cam.com/video/h-LnF1d6uJI/w-d-xo.html Im not a fan of 'hacky' code but whatever works I guess, Im glad you know this code does something about memory management at least! :)
@@TheCodingBug quote import is not a command and I went to path and then pasted to path to my appdata folder into to field to get the import command and nothing
► My Other Tutorials:
Build and Install OpenCV 4.5.1 With CUDA GPU Support on Windows 10
th-cam.com/video/YsmhKar8oOc/w-d-xo.html
Face Detection Using OpenCV Python with CUDA GPU Acceleration
th-cam.com/video/GXcy7Di1oys/w-d-xo.html
YOLOv4 On Android Using TFLite
th-cam.com/video/YzAjAS6Os8c/w-d-xo.html
Custom YOLOv4 Object Detection with TensorFlow and TFLite
th-cam.com/video/vzTCJM18uoM/w-d-xo.html
Darknet YOLOv4 Custom Object Detection: Part 2 (Training YOLOv4 Darknet)
th-cam.com/video/-NEB5P-SLi0/w-d-xo.html
Darknet YOLOv4 Custom Object Detection: Part 1 (Preparing Custom Dataset)
th-cam.com/video/sKDysNtnhJ4/w-d-xo.html
YOLOv4 Object Detection with TensorFlow, TFLite and TensorRT
th-cam.com/video/tCmC7nyfJp8/w-d-xo.html
Darknet YOLOv4 Object Detection for Windows 10 on Images, Videos, and Webcams
th-cam.com/video/FE2GBeKuqpc/w-d-xo.html
Real Time Object Detection on Webcam and Videos Using OpenCV With YOLOv3 and YOLOv4 | Windows Linux
th-cam.com/video/FjyF03uawsA/w-d-xo.html
Build and Install OpenCV 4.4.0 with CUDA (GPU) Support on Windows 10
th-cam.com/video/tjXkW0-4gME/w-d-xo.html
Install TensorFlow GPU and PyTorch with CUDA on Windows 10 Anaconda | CUDA 10.1 cuDNN 7.6
th-cam.com/video/PlW9zAg4cx8/w-d-xo.html
Real-time Multiple Object Tracking with YOLOv4 TensorFlow and Deep Sort | Linux, Windows
th-cam.com/video/GagII5PAeKg/w-d-xo.html
You made my fay
Second time i am going through this process. First time was a real hassle, this time knew where i should go directly. If you are watching this video to get the GPU running you are in the right place. Hands down the best tutorial ever.
After 3 hours of trying to install it, I finally found this video that worked.
Thank you very much :D
I am glad it was helpful.
Clean, simple with no excessive introduction.
and more importantly, it works!
Saved hours in just 9 minutes. Great content man !
You really are a lifesaver. Using both conda and pip had me worried, but it worked perfectly. Thank you.
wow nice this is the best tutorial i've ever seen, direct to the point tutorial, no blah blahs
"After 2 days of unsuccessful trials, your tutorial worked for me. Thanks a lot!"
been searching all across the web for hours and solved it with this video
I am glad it was helpful.
Very accurate and to the point information. After spending about 3 days, finally I found very useful video, hats off to you man
After 3 failed attempts, i found this video and it worked.
Thanks a lot man, it was so easy. Thanks a lot :-)
I am glad it was helpful!
Thank you so much! I had been struggling with this for two days. Shared it with my classmates to help them too!
I am glad it was helpful.
Bless you bro...spent wayyy too long messing with docker containers when this works literally first try
I am glad it was helpful.
After a weekend of thorough research i finally made it here and it works! Thanks so much!
I am glad it was helpful.
After 2 days of a painful trial and error and rage process, I finally managed to get it working. It is a shame the official documentation is not as clear as you were in your explanation. Thanks, bro!
I am glad it worked for you!
Clean and simple! Great work!
Thanks for the effort!
Dude, you seriously rock with this video. You literally made it look easy. Google themselves should pin this video to their stupid ugly-ass tutorial.
Thank you! finally I am able to make use of my Nvidea card a year after purchase!
Thank you bro! you literally saved me,
You earned a new subscriber.
You are a life saver!!!!! Thank you so much for this!!!!!!!!!!!!!!
helped me! for me to get it to work I still had to install cuda (11.2) and cudnn (8.1.1.) manually but the pointer to the tf webpage on building on win from source definitely helped; all peachy now, thank you - gonna go watch the openCV tutorial
I cannot believe my eyes. After 4 hours of tearing my hair out trying to mess with conda environments / path variables / version control only for it to not work. I come across this video, a glimmer of hope, a mirage in the desert... To my surprise as I approach the oasis in the desert, it does not vanish. Alas, it is real, it amazes me how horrible the official documentation and pay-to-view blog-posts are about this process. You have saved me my sanity and for that I am grateful, will be sharing this with my friends.
Now that my GPU is recognized now, Im getting an error: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize.
Any tips?
This usually happens because of mismatch TensorFlow and cuDNN versions. Make sure you're using the same versions as mine.
Sometimes, restarting the computer solves problem (if you have paths added to system variables, remove them).
Have you find the solution?
I did Tensorflow 2.7.0 which was condatoolkit=11.2 cudnn=8.1 and it worked in my python 3.9.7 environment.
with the same python environment, I tried to install the same but it didn't work. Don't know what the hell is happening!!
Bro you are legend !! 💪
Thank you sooo much. Was struggling with this.
Excellent, and it works! Thank you very much.
It works. Thanks!
Damn. Not sure why it only works this way nowadays but it worked.
Why would I have to use pip on a conda environment? Why would I have to use conda-forge instead of regular conda distribution? Why wouldn't the official anaconda instructions work properly (as they did a few months ago???)
Anyway, thanks dude. You saved my evening.
I am glad it worked for you.
Regular conda distributions miss the latest versions sometimes. At the time of making this video, it was not possible to use them.
Same is the case with pip. Only TF 2.4.0 was available with conda and v2.4.1 was available with pip.
Official instructions often lag behind the changes that are made for every new version.
@@TheCodingBug Yea but at some point I had conda installed TF-GPU 2.3.0, CUDA 10.1 and cuDNN 7.6 and TF STILL wouldn't recognize my GPU. Tried different configurations, none worked even though it should on paper... Also following the official Anaconda guideliness for TF-GPU installation doesnt work anymore. Then I did the install your way and it works like a charm. I would expect Tensorflow and Conda to have way better documentation if it comes at least to the INSTALLATION of the library. :P
This video is awesome. To the point.
I am glad it was helpful.
please do a video on installing TensorRT with cuda on windows
Great, thank you! I have been trying a lot of other tutorials and none worked, only yours. If i may ask, do i also need to install keras-gpu?
No you don't as keras now comes with TensorFlow (from tensorflow import keras)
Thank you. Fast and simple!
This has solved my problem. Thank you very much
It's good to hear that it was helpful.
Clean and simple! Thanks!!!
Perfect tutorial.. Thank you very much
Thank you for your help, this worked for me!!! Like!
tested with rtx 2060, tensorflow 2.2 cuda 10.1 and cuDNN 7.6
@@Jorge-wf3tg Thank you for sharing the configuration. Maybe it will be helpful to someone else.
You are the best. Can't thank you enough
Thanks it worked! btw when i trying to use vscode or install jupyter notebook it doesnt work. do you use this tool ?
I use it with virtual environment and then activate that environment in vs code. it works.
Good one 💪🏼
I’m getting error in import tensorflow as tf
Traceback (most recent call last)
File “”, line 1, in
ModuleNotFoundError: No module named ‘tensorflow’
Any idea from your side
you need to enable ipython kernel in your virtual environment
Am I to go through with yet another attempt at madness? Yes, I have been inspired.
Hope it works out!
Sir, I got trouble with this, and I got this error:
UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
and this :
2021-03-05 21:08:13.831889: E tensor flow/stream_executor/cuda/cuda_dnn.cc:340] Error retrieving driver version: Unimplemented: kernel reported driver version not implemented on Windows
and I don't know why i did exactly your procedure.
hi thank you so much for this! does this work on RTX 3090?
Thank you for the appreciation. I do not have an RTX card. But I am sure it will work. You can create a virtual environment and test it out. Do comment here if it works to help out others who might be wondering the same.
Crisp and clear
Thank you, you saved me
i've followed the steps my gpu is appearing in the command prompt but then when i try detect the gpu in jupyter notebook its not appearing?
You must've virtual environment, which needs to be activated in Jupyter notebook. Jupyter loads base environment by default.
@@TheCodingBug oh great thanks for reply! How can I activate the virtual environment in Jupiter notebook?
Can I install opencv gpu with this method?
No. OpenCV has to be built with CUDA support from scratch.
One question sir. I installed everything perfectly like you said in video & gpu is getting detected & working. However when I use gpu to train models, it only uses about 12% of gpu & improvement is not very significant over cpu. For comparison, I have R9 3900x & RTX 3080ti founder's edition. I ran a model once using CPU only & it takes 25s per epoch. If I use GPU, it takes 22s per epoch
Increase batch size of the model.
@@TheCodingBug I did it. Still not much improvement. Can you share your GPU vs CPU result? I ran a gpu tensorflow google colab program (first link on google search) on my jupyter notebook & it showed that gpu is roughly 31 times less time taking cpu on a random 2D conv test, but when I ran it on my own model, the results were not that significant.
@@yashdeepagrawal5572 same problem with me did you solve it?
@@rayansh1245 it seems that for certain models (mostly ANNs) there won’t be any significant improvement if you use GPUs even though on testing with tensorflow, my GPU was on an average 32x faster than GPU. So I think this is how it’s supposed to work. But do tell me if there is a way.
@@yashdeepagrawal5572 yeah similarly i have nvidia 3060 and i connected my jupyter notebook with gpu and still i see no improvement
THANK YOU! SUCH A SIMPLE PROCESS THAT IS MADE TO LOOK LIKE VERY HARD BY OTHER IDIOTS.
Thanks for the video. I am able to install Tensorflow using Intel UHD GPU instead of dedicated NVIDIA GTX 1650 GPU and that to only when I open Anaconda prompt as administrator. How could I utilise my Nvidia graphics card in this scenario? Any help would be really appreciated. Thank you.
@TheCodingBug Any help would be really helpful. I am stuck over here since a long time.
I didn't fully understand but if you have installed CUDA, cuDNN, latest Nvidia drivers, and TensorFlow GPU, it is going to pick nvidiw GPU, as shown at the end of this video. TensorFlow GPU does not utilize intel GPU.
I assume you're working on a laptop. Make sure you are using performance mode and not battery saving mode.
@@TheCodingBug thanks for the reply. Yes i had installed the things as you mentioned above and it was similar to the end of the video but later when I tried the same using jupyter notebook it showed me device:0 as the name of GPU used using the command tf.test.gpu_device_name() and the same device:0 was listed as INTEL GPU in my task manager when i was looking for utilization.
@@TheCodingBug and yes i will make sure I use the performance mode in my laptop.
Thanks a lot it worked for me finally
I am glad it was helpful.
After I try "import tensorflow as tf" to try it out, it tells me "ModuleNotFoundError: No module named 'tensorflow'" when I am sure I have followed all the steps properly. Please help me out :(
awesome, thank you so much :)
can we follow the same steps to install tensorflow-gpu=2.3.0 from conda instead?
Yes if it's still in conda repository.
Very nice, but I get:
Traceback (most recent call last):
File "C:\Users
atha\anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 64, in
from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal
What am I doing wrong?
Make sure you're working in a virtual environment with Python 3.8.
@@TheCodingBug I am, have realised it throws this error as my CPU does not have AVX support :/
Its not showing my GPU. I followed your instruction to the letter. I installed the same version of Visual C++ Redist, the same version of tensorflow and the same cudnn and cudatoolkit. Its still not working.
Same version of TensorFlow as well?
Hello, will u make a video "how to install YOLO with cuda?"
I already have made this tutorial: th-cam.com/video/FE2GBeKuqpc/w-d-xo.html
@@TheCodingBug tnx so much
Hello I am facing problem to using Premiere pro CC. Once I export my video and Unable to. Due to Error Compelling Move / GPU render error. Pls advice anything information needed to give me advice for solving the problem. Thanks in advance
@@TheCodingBug I am not clear the Answer. Please clarify the solution
You are my hero..
Is there a way to not install govnoconda on my pc?
I do the same as tutorial but i face this problem. Can you guide me to fix it.Thanks
2021-07-25 13:09:56.961151: W tensorflow/stream_executor/platform/default/dso_loader.cc:60] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found
2021-07-25 13:09:56.961262: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
You saved me. Thanks from the core of mine
I am glad I was able to help.
Thanks for your extremely helpful video. I don't have an appropriate graphics card on my home computer for tensorflow gpu, so I tried an Azure data science Windows virtual machine, which has a NVIDIA Tesla M60. Unfortunately, the tensorflow gpu version does not work on that Azure virtual machine. I upgraded to the latest Tesla M60 driver and I was able to follow your instructions to create a separate environment and install cudatoolkit, cudnn and tensorflow-gpu and get it to work.
Thank you for sharing. I hope this will be helpful for someone.
Sir, based on the Task manager it seems that my laptop is using CPU instead of GPU. Please help me. Thank you
Sir one dought is its necesary to install both tensorflow and tensorflow-gpu.
If any one knows then explain me
everything work so far no error but when I did the test on tf if it available on gpu it turn out false please help
It worked But i was training my network .GPU was not much affected(task manager)
. Have any idea??
THANK YOU!!
when i write this command no thing happen! "import tensorflow as tf "
Any help Please?
tf.test.is_gpu_available() false
Same problem
hi, I'm trying to use tf 1.14 for this and I notice that cudnn=7.4 can't be installed from any channel, I already tried the manual way of installing cuda, cudnn with the correct version stated in the tf version table it just doesn't work, any idea? probably not but thanks anyway
Can't install tensorflow 1 with this?? "Solving environment: failed with initial frozen solve. Retrying with flexible solve."
Doesn't work for me, got this error:
ImportError: DLL load failed while importing _pywrap_tensorflow_internal:..
I need Tf c++ installation any reference?? Pls help
showing error
(tf_gpu) PS C:\Windows\system32> pip install --upgrade tensorflow-gpu==2.4.1
Unable to create process using 'C:\Users\Ravi Chaurasiya\anaconda3\envs\tf_gpu\python.exe "C:\Users\Ravi Chaurasiya\anaconda3\envs\tf_gpu\Scripts\pip-script.py" install --upgrade tensorflow-gpu==2.4.1'
thanks mate
Thank you for such nice and easy way to install tensorflow_gpu video
i'm following your step and it works , gpu is detected by tensorflow . But when i start training the model i got the error with massage:
UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
[[node sequential_2/conv2d_6/Relu (defined at :1) ]] [Op:__inference_train_function_3151]
Function call stack:
train_function
any solution?
I get same error, would be nice if there was a fix
@@gunnerstone120 ok i've already got the solution.
There's a problem about memory management. U can insert this code :
import tensorflow as tf
config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.compat.v1.Session(config=config)
@@gunnerstone120 after that u must restart kernel and make sure u shut down other sessions
@@kikifade12 lol thanks for fast response, I was actually about to link you to this video that says same thing: th-cam.com/video/h-LnF1d6uJI/w-d-xo.html Im not a fan of 'hacky' code but whatever works I guess, Im glad you know this code does something about memory management at least! :)
@@gunnerstone120 hahaha np. tbh, i got the solution from the comment on that video
thanks a lot :)
can it work in jupyter?
Yes it will.
@@TheCodingBug yeay thank youuu
perfect
ModuleNotFoundError: No module named 'tensorflow'
You need to activate the environment you've installed TensorFlow in.
@@TheCodingBug I can't activate it on my visual studio code
S2 It works
Best till 2023
Hero
didnt work
It always works. Maybe you missed something.
@@TheCodingBug quote import is not a command and I went to path and then pasted to path to my appdata folder into to field to get the import command and nothing
You are a life saver man! Thanks a lot!!
I am glad you found it helpful.