Observing the Probabilities
What I observed while training a tiny language model: the loss stopped falling, rug's embedding never moved, and cat and dog were not as close as I expected.
Posts
The Assignment Was One JPEG
What it actually takes to run a 38 FPS model on a Jetson Nano, and why mine ran at 12.
Viewing Language Through a Probabilistic Lens
How next-word prediction, distributed representations, and a neural probability model let useful structure emerge from language.
How I Access My Bare-Metal Homelab When I Touch Grass
Why I put NetBird inside Talos to reach my bare-metal homelab without depending on Kubernetes or another always-on machine.
How I Made My Homelab Declarative
Migrating every hand-installed Helm release on my Talos cluster into git with Argo CD: the app-of-apps bootstrap, the adoption drill, and the surprises each component threw at me, until helm list -A came back empty.
Working With Text Data
How raw text becomes something an LLM can actually train on: tokenizing, building a vocabulary, byte pair encoding, sliding windows, and turning token IDs into token and positional embeddings.
Understanding LLMs
A foundational look at large language models, recurrent and convolutional architectures, transformers, and the difference between BERT and GPT.
What Open Source Has Given Me
A small note on open source, consistency, Meshery, and gratitude as my LFX term comes close to ending.
The Weight of More
A personal blog regarding my journey in open source and everything in between