Gradivex
  • Home
  • Academy
  • Arena
  • Visualizer
  • Models
Log in
NLP & Transformers/From Text to Tokens/Lesson 3

Byte-pair encoding

Module 1 · Lesson 3 of 5 · 7 min read
Theory

Nobody wrote the vocabulary

A natural assumption: some linguist at OpenAI curated the token list, choosing sensible prefixes and suffixes. Wrong. No human picked a single token. The vocabulary is learned by an algorithm from 1994 built for file compression, byte-pair encoding, and the recipe fits in one sentence.

Count every adjacent pair of symbols in your corpus, glue the most frequent pair into a new symbol, repeat. Start from single characters, run it 50,000 times, and the merge list IS the vocabulary. "t"+"h" merges early because "th" is everywhere. Later "th"+"e" merges into "the". Frequency decides everything.

This is a pattern you will keep meeting in this protocol: wherever you expect hand-crafted linguistic rules, you find brute statistics instead. BPE knows nothing about grammar, prefixes, or meaning. It knows what co-occurs, and that turns out to be enough.

Interactive

Run BPE by hand

A micro-corpus of eight words, already split into characters. The table shows the most frequent adjacent pairs right now. Merge your way to a vocabulary, one click per merge, exactly what the real algorithm does 50,000 times.

lowlowerlowestslowslowernewernewestwider
Most frequent adjacent pairs right now:
l+o ×5o+w ×5w+e ×5e+r ×4e+s ×2
Theory

From toy to production

The real numbers: GPT-2 ran BPE to 50,257 tokens. GPT-4o runs a descendant of it to roughly 200,000. Llama, Claude, Gemini: all BPE variants, different corpora, same merge-the-frequent-pair engine.

The corpus choice has consequences people rarely connect. BPE trained on mostly-English text spends its merge budget on English. The result:

  • "The cat is on the mat" → 6 tokens. Efficient, cheap, fast.
  • "Il gatto è sul tappeto", same meaning in Italian → roughly 9-10 tokens on GPT-2-era vocabularies.
  • Thai, Hindi, or Georgian text can cost 3-5x the tokens of equivalent English.

Same sentence, same model, but the Italian speaker pays more per thought, gets less effective context window, and waits longer per sentence. This is not a moral failing of the model; it is arithmetic falling out of a merge table. But it is the kind of arithmetic you should know exists before you deploy a product in twelve languages.

Interactive

Check your understanding

Why does "lower" tokenize into fewer pieces than "wixer", even though both are 5 letters?

Takeaway

What to remember

  • ✓BPE builds the vocabulary automatically: merge the most frequent adjacent pair, repeat ~50,000 times.
  • ✓It is a 1994 compression algorithm, not a linguistic theory. It rediscovers prefixes and suffixes from pure counting.
  • ✓The merge list is trained once, on a specific corpus, then frozen. The vocabulary is a snapshot of that corpus's statistics.
  • ✓Languages underrepresented in the corpus tokenize worse: more tokens for the same meaning, higher cost, smaller effective context.
  • ✓GPT-2: 50,257 tokens. GPT-4o: ~200,000. Different scale, same engine.
Mentor-0

Stuck on something in this lesson? Ask.

0/1000

Every lesson is free to read. Log in to save your progress, complete lessons, and earn Compute and XP.

Log in free
What is a token?Vocabulary and token IDs
Gradivex

Understand AI from the inside, read it, see it, prove it.

Platform

AcademyVisualizerRoadmapsModels

Community

DiscordLeaderboard

Company

PrivacyTerms

© 2026 Gradivex. All rights reserved.