LightGBM on SageMaker

It's possible to do, I have used it myself for gradient boosting, from within Jupyter you can simply run:

!conda install -y -c conda-forge lightgbm

Within a selected conda environment. No terminal access is needed, however it must be done, On the top right of the Jupyter notebook you can choose a terminal environment which will give you a shell to the backend instance and you can install there.

However if you want the notebook to be immutable/transferable you can do the install within the notebook .

Thanks