Kyle Pena articles - 7 total
Visualizing Decoder Layer Gradients
In this short post, I'll explain a practical problem you'll encounter when visualizing the gradients...
Gradient Descent on Token Input Embeddings
Input Embedding Space Gradients This is the first in a series of posts on the...
Investigating the performance of np.einsum
TLDR; tensordot mode (and therefore BLAS) is only activated for two-argument einsum when...
The Unreasonable Usefulness of numpy's einsum
Introduction I'd like to introduce you to the most useful method in Python,...
Use Hardware Security Modules And Browser Integration To Fight Deepfakes
Introduction Deepfakes are a massive problem without an effective solution. Current...
Capturing The Statistics of Streaming Data
The Rise (Return?) Of Specialized Computing Environments Isn't it interesting that the key...
Derivation of Welford's Algorithm
This will be somewhat out of context if you're coming here first. It's really a footnote in a much...