Spaced-repetition flashcards
Flashcards for Mathematics, Reconnected, Human Biology for the Medical Maker and the AI Acceleration primer — definitions, formulas, mastery checks and glossaries, each linked to its chapter — plus Big-O and coding interview decks that go with this site's tools — scheduled with FSRS (or SM-2), with cloze cards, your own decks, CSV and Anki import and export, and statistics, kept in your browser.
Added · updated 30 October 2026
Works with
- Big-O explorer and benchmark Formatting and referencerevise the costs the Big-O and data structures deck asks about
- Coding interviews Interview preppractise the patterns the coding interview deck covers
How to use it
Spaced repetition shows you a card just before you would forget it: a card you know comes back after days, then weeks, then months; one you miss comes back in minutes. A few minutes a day keeps a whole book in memory. The decks here are built from the three books on this site, and every card links back to the place in the book that teaches it.
How to use it
-
Study — pick a deck and a chapter, part or group (or All decks). The counts show the new cards left today, the cards in their learning steps and the reviews due. Read the card, recall the answer, then Show answer (Space, or tap the card) and grade yourself honestly: Again (forgot — it comes back in a few minutes), Hard, Good or Easy. Under each button is when the card will come back. Keys: Space shows the answer and then means Good, 1–4 grade, U undoes the last answer. On a phone, swipe the answered card: left again, down hard, right good, up easy. Suspend takes a card out of the queue (Browse brings it back). ⛶ Full screen (or F) shows just the card; Esc leaves. The answer ends with 📖 Read it in … — the section of the book the card came from, opened in a new tab (🛠 Open in … for the tool decks, which open the matching part of the tool).
-
Cloze cards hide a word or phrase in a sentence (
{{c1::…}}); each numbered deletion is reviewed separately. Short definitions in the maths deck are cloze cards: they ask for the term given its meaning. -
Decks — every book by part and chapter (and each tool deck by topic), with the cards, how many you have seen, how many are due and the share that is mature (an interval of three weeks or more). Study opens that chapter in the Study tab; 📖 opens it in the book (🛠 in the tool). Each deck can be downloaded as an Anki text file or CSV.
-
Browse — search the cards of the selection, filter by state (new, learning, review, due, suspended), open a card to see both sides, its FSRS stability and difficulty and its SM-2 ease, and suspend it or forget its progress.
-
My decks — make decks of your own: basic cards, basic-and-reverse (asked both ways) and cloze, with tags, an optional link and a live preview. Import CSV (a header row naming front, back, tags, type and link — or just two columns) or Anki’s Notes in Plain Text export (tab separated; its
#separator:,#html:and#tags column:headers are read, HTML is turned into the card markup and MathJax\( \)into$ $). Export a deck as CSV, Anki text or JSON. -
Statistics — for the selection: cards new, learning, young and mature; true retention over 7, 30 and 90 days (the share of reviews of graduated cards you did not answer Again); the mean predicted recall right now; a 30-day forecast of reviews due; reviews per day over the last month; and a table by chapter.
-
Settings — the scheduler (FSRS-6, FSRS-5 or SM-2), the desired retention, learning and relearning steps, the longest interval, interval fuzz, new cards and reviews per day (per deck), the order of new cards, the hour the study day starts, your own FSRS parameters, and the backup: export or restore everything as one JSON file.
Card markup, for your own decks: bold, italic, , maths as code $…$ and …
(typeset with KaTeX; write \$ for a dollar sign), lines starting - ` or `1. ` for lists, and
`[text](https://…) for links.
The decks
-
Mathematics, Reconnected — about 560 cards across the nine parts: the book’s Refresher boxes (definitions), the cheat-sheet rules and formulas, every chapter’s mastery-check questions with the book’s answers, and recognition exercises. A mastery question links to the section its decision rule says to reread. To practise the skills rather than recall them, use the maths practice gym, which generates problems and schedules each skill with SM-2.
-
Human Biology for the Medical Maker — about 360 cards: each chapter’s questions with the book’s answers and reasoning, and terms from the working glossary, linked to the section where the term is introduced.
-
AI Acceleration — about 390 cards: the book’s glossary of terms and acronyms and its numbers worth remembering (each linked to the Part and section that teaches it), and about 280 cards written by hand from every Part, each linked to its section of the primer.
-
Big-O and data structures — about 145 cards written for this site: growth classes and the definitions of O, Ω and Θ; costing loops and spotting hidden costs; recurrences and the master theorem; amortised analysis; the costs of operations on arrays, lists, hash tables, trees, heaps, tries, union-find and graphs; and sorting, searching and graph algorithms. Each card opens the matching tab of the Big-O explorer (the cheat sheet, the revision quizzes, the recursion trees, the amortised append and so on).
-
Coding interview patterns — about 220 cards written for this site, one chapter for each of the thirty-three patterns of the coding interview trainer (the signals that point to it, its key steps, costs and pitfalls) and one on the interview itself: clarifying, testing, stating complexity and STAR stories. Each card opens its pattern in the trainer.
The book decks are generated from the books' sources by scripts/flashcards-gen.mjs and curated by
hand; the books themselves are not changed. The two tool decks are written by hand
(flashcards-bigo.js, flashcards-interview.js). Your progress is stored only in this browser
(IndexedDB) and never leaves it; export a backup to keep it or move it to another browser.
How the scheduling works
FSRS (Free Spaced Repetition Scheduler, by Jarrett Ye and the open-spaced-repetition project — the algorithm Anki has offered since 23.10) models each card’s memory by its stability S (the interval at which you would still recall it with 90 % probability) and difficulty D (1–10). The probability of recall t days after a review is R = (1 + F·t/S)−w20 with F chosen so that R = 90 % at t = S (FSRS-5 fixes the decay at 0.5; FSRS-6 learns it). A successful review multiplies S by a factor that is larger for easy cards, for small S, and for a card reviewed when R had dropped low; a lapse resets S to a fraction of its old value. The next interval is the t at which R falls to your desired retention. The formulas and the default parameters are those published in The Algorithm (FSRS-5: 19 parameters; FSRS-6: 21), and the learning steps, same-day reviews and rounding follow the reference scheduler py-fsrs; the tests check this tool against py-fsrs’s own published test values.
SM-2 is the 1990 SuperMemo algorithm (P. A. Wozniak, Optimization of learning): intervals of 1 and 6 days, then the last interval times an ease factor EF (starting at 2.5, rounded up), with EF' = EF + 0.1 − (5 − q)(0.08 + (5 − q)·0.02), never below 1.3. The buttons map to q = 1 (again), 3 (hard), 4 (good) and 5 (easy); a fail restarts the repetitions and shows the card again after the first relearning step. Both schedules are kept for every card, so you can switch between them.
Limits
FSRS uses its published default parameters unless you paste your own (for example from Anki’s optimizer); this tool does not fit parameters to your review history. Retrievability uses whole days since the last review, as py-fsrs does. Daily limits count per deck and reset at the hour you choose; "Learn 10 more" raises them for this session only. Card text is a small markup, not HTML — Anki imports lose images, audio and colours. A card’s link points at the published book; if the book is reorganised, a link may land on the chapter rather than the section until the decks are regenerated. Grading is self-assessment: the scheduler is only as honest as your buttons. Everything lives in one browser: clearing site data deletes it, so export a backup now and then.
Related: maths practice gym, Big-O explorer, coding interviews, algebra step by step, symbolic algebra, AI hardware calculator.
Companion to Mathematics, Reconnected, Human Biology for the Medical Maker and AI Acceleration.
Linking to an example
Share link (under the title) copies a URL that opens the tool on the same deck and chapter. Your progress, settings and own cards are never included.
You can also write a link by hand. Add the tab after #: study, decks, browse, mine,
stats or settings (a link with a deck opens Study).
| Parameter | Meaning | Example |
|---|---|---|
|
|
|
|
A chapter ( |
|
|
Only one kind of card: |
|
|
A preset (below) |
|
For example ?deck=math&chapter=ch18 studies
chapter 18 of Mathematics, Reconnected, and
?deck=math&chapter=part5&kind=formula
only the calculus formulas.
Groups: Mathematics, Reconnected part1 (Foundations) to part9 (Numerical mathematics and
Horizons); Human Biology body (ch. 1–4), structure (5–8), systems (9–12), devices
(13–17), practice (18–21); AI Acceleration foundations (Parts I–II), memory (III–V), silicon
(VI–IX), machine (X–XII), power (XIII–XV), scale (XVI–XVIII), frontier (XIX–XXV).
Big-O and data structures analysis (growth, loops, recursion, amortised), structures
(arrays and lists, hash tables, trees and heaps, tries and union-find), algorithms (sorting,
searching and graphs), with the chapters growth, loops, recursion, amortised, arrays,
hashing, trees, special, sorting and graphs; Coding interview patterns arrays-group,
order-group (heaps, intervals, linked lists, bits), graphs-group (trees, graphs, union-find,
tries), search-group (backtracking, dynamic programming, greedy) and meta (the interview
itself), whose chapters are the trainer’s pattern ids (two-pointers … bit-manipulation) and
process.
example= |
Opens |
|---|---|
|
Mathematics, Reconnected chapter 1: numbers, the number line and arithmetic |
|
chapter 8: quadratics |
|
chapters 17–18: the derivative and its uses |
|
Part V: calculus (chapters 16–20) |
|
Part VII: linear algebra (vectors, matrices, eigenvalues) |
|
Part VIII: probability and statistics |
|
Human Biology chapter 9: circulation, breathing and the wider body |
|
chapters 1–4: anatomy, chemistry and cells |
|
chapters 13–17: materials, making, measurement and devices |
|
AI Acceleration Parts I–II: why hardware changed, data encoding |
|
the serving Part: batching, quantization, speculative decoding |
|
the numbers worth remembering (all Parts) |
|
Big-O and data structures: the whole deck |
|
growth classes and how to analyse loops, recursion and amortised costs |
|
amortised analysis: dynamic arrays, two-stack queues, union-find |
|
operation costs of arrays, lists, hash tables, trees, heaps, tries and graphs |
|
Coding interview patterns: the whole deck |
|
dynamic programming |
|
trees, graphs, union-find and tries |
|
the interview itself: clarifying, testing, complexity and STAR stories |
Feedback is a gift. Something to improve, a feature to add, or a bug? Tell me.
Runs entirely in your browser; nothing you type leaves this page.