A Flutter and Python project to visually educate users how to build AI using TensorFlow and Keras. Please make sure you have around 5GB of space available on your device. (Eventually you will need more, perhaps another 1 or 2 GB for the AI)
Note that even if you are only working on Flutter, you still need to setup python even if you don't write code in it.
Python devs also need to setup Flutter so they can test their backend changes through the UI (as well as testing through the code).
Please ensure that git is installed and you can manage git branches. For VSCode I like the Git Graph and GitLens extensions in tandem with the normal GitHub extensions.
Follow the instructions on https://docs.flutter.dev/get-started/install (Feel free to skip the setup for Android and Web sections) for your system and then clone this repository into a directory of your choosing. Note that around 2.5 GB will be taken up by flutter and this repository is about 600MB at the time I wrote this -- minus the python libraries.
Verify that you have setup your environment correctly by running the application. (Note that the build process can take up to a minute and will most definitely take longer the first time you run the app)
Ignore any errors you may see about Python if you haven't setup python yet.
Setup a Python 3.9.10 venv (virtual environment) in your workspace by running python -m venv .venv in your workspace. Obviously, install 3.9.10 if you haven't done so already.
If you are on Windows and VS Code, close all your terminals and open a new terminal. If an error appears, open a PowerShell window in Administrator mode and execute the following command: Set-ExecutionPolicy RemoteSigned. Then reopen a terminal window.
Run python -m pip install --upgrade pip. This is required, it doesn't just make it look cooler.
Then run pip install -r requirements.txt. (Note that around 1.4GB of space will be taken up here).
This will take care of most of the heavy lifting for you.
If you have a compatible Nvidia (Steps 2 and 3) or AMD GPU, please go to the respective website and install the components necessary for Python to leverage your GPU in AI calculations. Note that this is not necessary but it will be beneficial if you are creating/testing anything beyond a very small model.
Eventually we will want to natively detect if the system running our app has a TensorFlow compatible GPU for the AI and redirect the user to the appropriate website to install those components.
I don't expect everyone to start sprinting. I know a lot of these things we are working with will be new to you. Don't feel pressured to deliver items just because I have an endless list of to-dos. I would much rather have quality code than lots of broken code. That being said, don't slack off. Learn as much as you can and try to contribute to the project as soon as you can and in whatever way you can.
Python
- Pydantic is how python stores and validates the data that comes and goes from Dart (the frontend).
- Keras/TensorFlow is the AI library we will be using. We will be using the functional API heavily.
Flutter
- Dart is the programming language of Flutter, master it like the Force. You can play with dart here: https://dartpad.dev/
- Basic Flutter: A great way to learn the basic principles of Flutter.
- We will need to use actual assets instead of the placeholder images I've been using.
Pretty much the only rule I have: DO NOT COMMIT TO THE MASTER BRANCH!!!
Before you type anything make sure that you are in another branch! When you feel that you are finished working on whatever change you are working on: DO NOT JUST MERGE THE BRANCH INTO MASTER!!!
Create a Pull Request (ask me or Google if you don't know how) and notify me. I will look at your changes and either ask for additional changes or approve the request and pull it into master.
Above all, if you get stuck or have a question, please ask myself or another of our teammates for help. It doesn't help anyone if you're just staring at text on a screen and have no idea what's going on. I wrote all the code before May 13, so I (most likely) know what it does!
- Ask me and I'll give you something to do. Certain things need to be done before other things. If it's on the board it's a safe bet you can work on it. I can write new cards if needed.
- Look at the Trello board and if you see a bug or feature you want to work on that someone else is not working on, feel free to assign it to yourself and begin work. If you would like to work on a ticket with someone specifically, ask that person first before assigning yourself to the card. Common courtesy stuff.
If you need to change a schema in python, notify myself and those working on the frontend. They need to know that the data you are sending them is going to change. If you need to add/remove/change something in a schema and you are on the frontend team, notify myself and we will make a card in Trello so someone on the backend can work on it (which I will probably just do anyway to expedite the process).
They are autogenerated for us based on the schemas the backend has created.
If you are working in python and you change a schema or enum in any way, you must then run the following command:
python pydantic_dart_converter.py
This will first take the python schemas and create/overwrite their dart equivalents. Then it will take those newly created dart schemas and flesh them out for use by the frontend team.