Machine Learning
-
CatBoost stands out by directly tackling a long-standing challenge in gradient boosting—how to handle categorical…
10 min read -
Reverse-engineering large languages models’ computation circuit to understand their decision-making processes
7 min read -
Understanding all versions of flash attention through a triton implementation
16 min read -
Optimizing highly parallel AI algorithm execution
11 min read -
How attention helped models like RNNs mitigate the vanishing gradient problem and capture long-range dependencies…
10 min read -
One year later: what I learned still matters
19 min read -
The ML Uncertainty Package
15 min read -
Neural networks under a different lens: generating basins of attraction in a shift register NN
12 min read -
Breaking down my role as a machine learning engineer
8 min read -
A pragmatic look into protecting algorithms and models deployed into real-world federated analysis and learning…
11 min read