Context: I got this error when running tensorflow with python on Ubuntu 20.04, after installing the cuda toolkit by choosing the local deb installation file.
Solution: the solution was to download and install the deb network file instead
If you run tensorflow in a docker console you can run the following code in your terminal:
docker exec b61a27999cb8 python3 -c "import tensorflow as tf; tf.compat.v1.Session(config=tf.compat.v1.ConfigProto(log_device_placement=True))"
If you run a jupyter notebook you add the following line:
import tensorflow as tf
print("Num GPUs Available: ", len(tf.config.list_physical_devices('GPU')))
https://docs.docker.com/storage/
https://docs.docker.com/storage/volumes/
docker run -it -p 8888:8888 -v /home/personal/Documents/tensorflowData:/tf/notebooks tensorflow/tensorflow:latest-gpu-jupyter
Change SOURCE_FOLDER and DEST_FOLDER accordingly (use absolute paths!).
Now if you navigate to localhost:8888 and create a notebook on DEST_FOLDER, it also should be available on SOURCE_FOLDER.
For tensorflow 2, the destination folder is /tf. You can create a folder of your choice, say: notebook, where you will save your file in your jupyter notebook. Those files will be available in the SOURCE_FOLDER you have set.
The changes in tensorflow-tutorials won’t be kept even if you choose it as your destination folder. You have to use another folder as detailed above.