Learning Notes

My personal knowledge base — distilling what I read, learn, and debug into searchable notes.

• Machine Learning

  ---

  Models, training, tuning, and lessons learned.

  Enter

• Robotics

  ---

  Robot learning / control / perception / simulation.

  Enter

• Mathematics

  ---

  Linear algebra / probability / optimization fundamentals.

  Enter

• Reading

  ---

  Notes on books / blogs / courses. (Coming soon)

  Enter

All notes

• Action Tokenization: Discretizing Continuous Actions into Tokens Robotics · 2026-08-06 — Unpacking every link in the chain "continuous action → discrete token → continuous action", along the π₀-FAST-style discrete autoregressive route.
• Flow Matching ODE Solvers, Illustrated Machine Learning · 2026-08-06 — The accuracy-cost trade-offs of Euler / Midpoint / Heun / RK4 / DoPri5 in flow-matching sampling, with interactive experiments.
• Autoregressive Models: BERT / GPT and Prefix-LM Architectures Explained Machine Learning · 2026-08-06 — How the sequence-modelling paradigms differ, and why robot VLAs use a Prefix-LM hybrid.
• Fourier Transform & DCT: From Frequency to Energy Compaction Mathematics · 2026-08-06 — From Fourier series to the DCT's energy compaction: why a smooth signal's energy piles up on a few low-frequency coefficients.
• Lie Theory: An Intuitive Introduction for Optimization and Robotics Mathematics · 2026-08-06 — Manifolds, tangent spaces, and the ⊞/⊟ operators: why optimization over rotations and poses works the way it does.

---

Style Guide — the reference sheet for the site's shared styles.