Technical books and Kindle reads. Each review covers what the book argues, what I learned, how I applied it in my own work, and which of its references are worth following.
From autoregressive models to VAEs, flows and diffusion, each built from probability theory and code.
Reading
Four phases for any problem: understand it, make a plan, carry it out, look back.

Retrieval, spacing and interleaving beat rereading, even when they feel slower.
Reading
The science of memory: why forgetting helps, and how to use spacing, testing and context.
ReadingThe brain's four pillars of learning: attention, active engagement, error feedback and consolidation.
Reading
Learning research turned into classroom practice: retrieval, spacing, interleaving and feedback.
ReadingA consensus report on what research says about how people learn, from infancy to adulthood.
Reading
Eight strategies that make learners build understanding: summarising, mapping, drawing, imagining, self-testing, self-explaining, teaching and enacting.
Reading
How to turn a topic into a question, a question into an argument, and an argument into a report.

Clear nonfiction comes from cutting clutter until only the thought is left.

Half memoir, half toolbox: read a lot, write a lot, and cut the second draft by a tenth.

Rules of usage and composition, summed up in three words: omit needless words.