Sagemaker Kernel for tensorflow and pymysql

Question

I am reading from a database using pymysql while using conda_python3 kernel but, when I want to use tensorflow/keras for deep learning, notebook instance doesn't recognize it.

I tried installing keras and tensorflow libraries but after showing a success message and restarting kernals, I got the same old error that "No module named 'keras'".

Reading stackoverflow related posts, I was able to temporarily solve the issue with switching to another kernel "conda_tensorflow_p36".

Do you know any way to avoid switching between these kernels (maybe a better way of installing packages in sagemaker notebook instances?).

Answer 1

I found the answer and sharing it here,

Don't install anything in the notebook itself and use the terminal instead,

For the installation you should specify the name of the package and the kernel that you want the package to be installed. For example,

conda install -n tensorflow_p36 -c conda-forge pymysql