Understanding Attention Mechanisms in Transformers
A deep dive into how self-attention works, why it replaced recurrent architectures, and the mathematical intuition behind scaled dot-product attention.
Writing about mathematics, data science, and artificial intelligence.
A deep dive into how self-attention works, why it replaced recurrent architectures, and the mathematical intuition behind scaled dot-product attention.
Exploring the fundamentals of Bayesian thinking and how to apply prior knowledge to update beliefs with evidence — with Python examples.
Why feature engineering still matters in the era of deep learning, and practical techniques that consistently improve model performance across domains.
Breaking down gradient descent, momentum, Adam, and the geometry of optimisation landscapes — with interactive visualisations.