An amalgam of how-to-use guides on software tools, explanations of topics I found interesting and some personal thoughts and views. I would love to hear your comments/suggestions on these topics. Please reach out via email !
-
Normalizing Flows
These are a class of generative models that are often cited as flow models as they leverage the change-of-variables formula from probability theory to transform samples from a simple prior distribution, to model the desired data distribution, via a sequence of invertible transformations.
-
Denoising Diffusion Probabilistic Models
In this blogpost I dive into a popular class of generative models called diffusion probabilistic models.
-
Searching for evolutionarily conserved sequences using Multiple Sequence Alignment (MSA)
In this blogpost I explain how Multiple Sequence Alignment (MSA) works using a toy example and how two different classes of algorithms implement MSA in practice.
-
AlphaFold3 : Predicting Biomolecular Structures from Sequence
AlphaFold3 represents Google DeepMind's as of date latest open source model for bio-molecular structure prediction. In this tech blog, I dive deep into how AlphaFold3 works, how does it perform on benchmark datasets and what are its limitations.
-
Regularize your models ! Part 1. Parameter Norm Penalties
In this tutorial we will look at a widely used strategy to learn sparse model parameters by adding parameter norm penalties to the loss function.
-
Creating python packages using build and setuptools
You have done the hard work ! Now make it easy for others to use and build on your code by creating a Python package. Once done, your code is just a 'pip install' away from finding a new home !
-
PyTorch vs TensorFlow: A comparison
Alright, so now you know the basics of how neural networks work. You might be wondering which python framework would be best for you to start building your own nets. This interactive google collab notebook will help you choose between the two most popular frameworks: PyTorch and TensorFlow.