The Cost of Being Wrong: KL Divergence
If Cross-Entropy is the total cost of sending a message, KL Divergence is the unnecessary tax you pay for having a bad model.
BLOG
Aricles, Stories, and Thoughts
This blog mainly serves as a record of what I learn and a reference for myself. I document the mistakes, screw-ups, and little (some huge) discoveries I make while coding and building things.
Filter by
All Posts
Capture Activations and Attention Maps using PyTorch Hooks
Hooks allow us to interrupt the dynamic computational graph created by PyTorch in the forward or backward pass.
From Telecom Lectures to Deep Learning: Finally Understanding Entropy
When we train neural networks, we are essentially trying to lower our surprise about the data until our model's predictions align perfectly with reality.
A guide for the unkown: #ScholarX
Enhancing Low Light Images: A Deep Dive into Autoencoders
Hybrid Images through Frequency Wizardry
Flower Image Classifier
Flower Image Classification using Transfer Learning