Choosing a model and judging your own application are two different evaluation problems. Benchmarks answer the first. Error analysis, graders and a small eval set answer the second.

A map of the series on Scientific Spaces (科学空间) and a reading order for the math behind machine learning.

Most of us learned to ask questions in a room where the teacher already held the answer. Why that habit follows us out of the exam hall, and what breaks it.

Sebastian Raschka's blog has about 190 posts over thirteen years. A reading order that moves from method to map to practice to the current frontier.

A stage-by-stage map of the best resources for learning LLMs, from foundational math to research-level papers.

VHF radio procedure, sound and light signals, and why marine communication is built around strict, unambiguous formats rather than natural conversation.

The emergency procedures Day Skipper drills hardest, and why the course treats rehearsed response over improvised heroics.

The handful of knots and the underlying principles that make docking and anchoring predictable instead of improvised.

How the course ties together three separate moving systems — wind, tide, and current — into the single practical question of what the water will actually do to your boat.

How marine radar builds a picture out of nothing you can see, and the discipline needed to trust it in fog, at night, or in busy traffic.