Some open-source software that I maintain.
This Python package is a PyTorch framework for digging into the inner workings of hierarchical visual models (e.g. deep nets). The part of the package to which I contribute most is
synthesize/eigendistortion and models in the
simulate branch. It is a fun mix of machine learning and numerical linear algebra.
I started learning Python in 2015 by working through the textbook “Analyzing neural time series data” by Mike Cohen, and translating all of its code from MATLAB. There’s a ton of nice signal processing tools in here, and to my surprise it now has over 120 stars on GitHub.