cbhyphen – cbhyphen.github.io

Maximum Subarray Sum

It’s been a few months since I posted due to health issues, but I’m breaking the silence with something small today. Most of my posts have been deep dives so this will be a…

Speeding up Inference with TensorRT

follow up post to pytorch quantization … can we make it faster with GPU and TensorRT

Sample Size for A/B Hypothesis Testing

The other day I had a conversation at work about model training and evaluation metrics. A colleague of mine suggested that the small improvements we were seeing might not…

Matrix Factorization for Recommender Systems

It’s pretty difficult to escape recommender systems in 2024. These days, we regularly get recommended things such as content from streaming services like Netflix and…

FX Graph Mode Quantization in PyTorch

Following up on the last post I did about eager mode quantization in pytorch, in this post I’ll be using pytorch’s FX graph mode quantization to quantize the same R-CNN.…

Eager Mode Quantization in PyTorch: Quantizing the Backbone of an Object Detection Model

Model quantization is a powerful tool to reduce memory and compute by using lower precision data types. Although I’m familiar with the concept and a few high-level APIs, I…

Gaussian Bayesian Classifiers

Even though I studied (and revisited, and revisited..) Bayesian statistics several times over the years, I’ve always felt that it wasn’t an intuitive paradigm. Specifically…