This is where I store what insights in ML and software dev I came across and I felt compelled to properly structure my thought so I could at least refer that to colleagues when poorly trying to explain them on the fly.

Posts

• What tokenization achieves

  Going beyond transformers

  Jan 14, 2024

• Deploying a snapshat filter to the browser

  How to deploy a ML model without worrying about a cloud pipeline.

  Jul 29, 2020

• Creating a translation app

  How to create a custom clone of translate.google.com

  Jul 27, 2020

• Super simple estimation of available solar energy

  How much energy is actually renewable and humanity's cap in consumption

  Mar 19, 2020

• Can we train neural networks without gradient descent ?

  If the lottery ticket hypothesis is real, does that mean we can train a neural network without gradient descent?

  Mar 10, 2020

• Running a docker with GPU enabled (for pytorch and tensorflow)

  Mar 4, 2020

• Self KL-divergence for detecting out of distribution data and unsupervised text classification

  Running two models alongside each other for trivial out of distribution detection in production models and side bonus is getting unsupervised text classification

  Feb 26, 2020

• Model based encodings (3)

  Aug 6, 2019

• Model based encodings (2)

  Jun 6, 2019

• Model based encodings

  How to use BPE without this hardcoded algorithm

  May 16, 2019

• A third of your math work is useless

  AI predicts confidently how students will do.

  Apr 13, 2018

• A non-NLP application of Word2Vec

  Applying Word2Vec to math exam questions.

  Jul 31, 2017