I am looking for a way in which I can write my code in a terminal environment (in Vim or Emacs) and yet (temporarily) use Google Colab's hardware resources (like a GPU or TPU) to train a machine learning model. I am not sure what is a good way to go about this.
EDIT: This is not meant to trigger opinionated answers. I am just trying to understand how it could be done at a high level to understand the options.
Here's a high level sketch of what I am thinking. However, please excuse any misconceptions or poor design apparent in the diagram -- that is, this image is just to help illustrate my question, not part my requirements. The only real requirements I have are integrating a terminal based workflow with Colab GPU / TPU resources. It's an open ended question, I'm just trying to learn what are the options of how something like this could be done, thanks in advance.
