16 July 2026
4 min read
Building a tiny language model from scratch
Build a decoder-only Transformer one understandable component at a time, train it on a laptop, and generate text from your own checkpoint.
LLMsAIPyTorchTransformers
Articles
Long-form notes and essays. Mostly practical, occasionally speculative, and designed to leave enough room for code and working demos.
16 July 2026
4 min read
Build a decoder-only Transformer one understandable component at a time, train it on a laptop, and generate text from your own checkpoint.
27 May 2026
14 min read
In an era where we all talk about, and have our feeds full of intelligence, why can we still not define it?
26 May 2026
12 min read
Why most teams should begin with hosted models, owned telemetry and open-weight experiments before considering pretraining from scratch.