GMM showed me soft clustering beats hard boundaries for messy data. Gradient boosting is just trees fixing each other's mistakes, and writing backprop in NumPy? pure clarity.
Stop calling fit(). Build it.
I have hosted a separate website for the implementations,
and I will add more paper-based implementations once I'm done with DL basics.
Here's the site: http://papers.mahirmalik.in
Also, the GitHub repo: https://github.com/mahirmlk/papers-implementations.git
Top comments (0)