- information-theory
- language-modeling
- pytorch
- teaching
- vision
- work
•
•
•
•
•
-
Principal component analysis of BERT's static embeddings
Principal component analysis (PCA) is a simple algorithm that compresses high-dimensional vectors into a low-dimensional space. It does this by creating an orthogonal basis in the low-dimensional space using the directions along which the data has maximum variance in the original high-dimensional space.
-
Creating a distribution with a specific entropy using PyTorch
Using an optimizer to play with the statistical properties of a distribution.
-
Lecture on n-grams
Guest lecture on n-grams delivered virtually at Walter Payton College Prepatory High School in Chicago, IL.
-
LaTeX workshop
Slides for a workshop I gave on general LaTeX usage.
-
Training a CNN on the MNIST dataset
Code that trains a convolutional neural network to recognize hand-written digits. This notebook is adapted from some guest lectures I gave for courses for which I served as a TA.
-
Getting started with PyTorch in Google Colab
A brief introduction to using PyTorch in Colab. This notebook is adapted from some guest lectures I gave for courses for which I served as a TA.
-
Shannon information
A brief introduction to Shannon information and its applications to language and language modelings. I put this notebook together as my contribution to a seminar in which every student was required to present on a technical topic.