Drill the concepts that separate practitioners from users
These fifty flashcards cover the advanced AI and machine-learning vocabulary that ML engineers and researchers are expected to know cold — parameter-efficient fine-tuning, attention and positional-encoding variants, decoding and serving optimisations, alignment methods, and the architectural tricks behind modern large models. They assume you already understand the basics and want to sharpen the harder material into instant recall, the kind you need in an interview or a design discussion. Every card pairs a precise term with a concise, technically accurate definition that tells you what it is and when it matters.
How it works
Pick a category from the dropdown to filter the deck — fine-tuning, architecture, inference, alignment, or all of them at once — then work through the cards one at a time. Each card shows a term; before you flip it, try to state the definition in your own words. Tap to reveal the answer, compare it to your recall, and advance. The deck cycles, so you can loop through repeatedly and let the cards you miss surface again.
Tips for getting the most from them
Active recall is the whole point: the value comes from the effort of retrieving a definition before you see it, not from reading the answer. Say it out loud or write it down, then check. Space your reviews across several days rather than cramming once, because spaced repetition is what moves these concepts into durable memory. Filter to your weakest category and drill it until those cards feel easy, then widen back out to All for a full pass. If a card keeps tripping you up, go read the source paper behind it — the flashcard is the index, the paper is the depth.