meanderingstream / dl_foundations_in_elixir

Deep Learning from the Foundations with Elixir. This repository is a transformation of the 2022 version of Fast.ai's Deep Learning for Coders Part 2 into Elixir. These code notebooks follow the notebooks from the [first portion](https://github.com/fastai/course22p2) of the course.

Deep Learning from the Foundations with Elixir

This repository is a transformation of the 2022 version of Fast.ai's Deep Learning for Coders Part 2. These code notebooks follow the notebooks from the first portion of the course. Please watch the Jeremy Howard's course videos for the complete context. In the course, Jeremy starts with some constraints. His approach is to first implement a concept in standard Python code. Once he's introduced the concepts, then the notebooks start to bring in PyTorch library code that also implements the concept. He tries to incrementally demystify the PyTorch and Fast.ai library code.

Similar to the course, these notebooks start from standard Elixir code and then bring in Nx and Axon libraries. We'll use Elixir and Livebook.dev interactive & collaborative code notebooks. A requirement for these notebooks is a running Livebook application or server. Livebook runs on Windows and Mac desktops and on Linux. Please see the Livebook web site for instructions on installing the basic Livebook application. Livebook also runs on Linux. For our purposes, we run Livebook on a local Linux server using escript. For more information on using Livebook on escript, please see the Readme.md at https://github.com/livebook-dev/livebook.

We'll be building a livebook for every Fast.ai Foundations Jupyter notebook. We welcome pull requests that improve our notebooks.

dl_foundations_in_elixir's Issues

Example of showing normalized 2D data in a Livebook

Problem
I'm not sure how to show a black and white image for normalized Float values between 0 and 1. Also need a solution for showing 2D Nx.Tensor between 0 and 1.

Solution Approach
Add a pull request to 01_matmul or a snippet example that can be used to replace the TODO section on showing a 2D list of list of Float numbers.

The best approach would be compatible with Windows servers in addition to Linux and Macs.

Ideas
Convert normalized values by multiplying by 255 and turn into Integer numbers. Load the data into Kino.Image.

The Image library might be helpful here or maybe eVision.

Create a guide or video on how to setup a Livebook server on a cloud GPU server

Starting a GPU capable Livebook.dev server
It can be hard for users to start with DL in Elixir because they don't have easy to repeat examples of getting access to a GPU. The easiest way to get started with Nx/Axon is a free or low-cost cloud GPU service. How can a user transform a Python focused service to run livebooks?

The biggest problem with using cloud GPU services is their focus on Python. I run a local Livebook Ubuntu server. I don't have any experience in setting up Livebook on a cloud service.
Solution Approach
Create a persistent, free, guide that helps Elixir folks run on GPU cloud services economically. We'd like to have a page of links to blog posts or videos that describe how to set up a cloud GPU Livebook. Alternative, we would be pleased to add a page in this repository that contains a guide.

Potential Cloud Services

• Paperspace Gradient
  I believe that Paperspace has the easiest persistent disk approach and a free tier.
• Google Collab
  For those with a Google Cloud drive, this might be a good option. I know that my son had good success using Collab for PyTorch/Fast.ai
• Kaggle
  Many of the 2022 Fast.ai courses were originally run on Kaggle.