Sandbox Saturday: Codes, Papers, Tutorials, ...
Hands-on things to tinker with over your weekend to upskill.
A curated list of hands-on resources you can jump right into and ‘learn by doing’. Goodluck and have fun!
This post covers:
MATLAB on Google Colab
PINNs from scratch code
ML in fluid dynamics overview
Free NVIDIA education
Science newsletters
1. MATLAB on Google Colab
I have been pretty vocal about my love for Google Colab. First and foremost, the free access to compute (GPUs) is hard to beat. Further, for only a few bucks a month, you get access to a higher tier membership with even better GPUs. For example, I pay $10/month there and this is the VRAM I get 😈 (do you know how much that would cost to buy for your house desktop machine?).
This hands-on tutorial (🔗 link here) is quite fun- how you can slot MATLAB into this Google Colab sandbox. Specifically, this tutorial shows how you can train an AI model and export it to TensorFlow (or ONNX) to call it from Python.
The tutorial is a walkthrough that turns ends with a complete deep learning workflow. At a high level, the steps covered are:
1. Connect your notebook to a runtime with a GPU
2. Open the terminal to install MATLAB in two lines
3. Add Deep Learning Toolbox (and friends)
4. Launch MATLAB (again) & Verify the GPU
5. Hands-on Example: Time-Series Forecasting with an LSTM
6. Export the network to TensorFlow
7. Round-trip test in Python
8. Gotchas & pro tips
Enjoy!