From Claude Shannon's 1948 entropy equations to LLM-based translation — every system in this course is implemented in Python, runnable in your browser, and tested on the same 20 sentences so you can watch quality evolve across eras.
Hover a milestone for context. Click to jump to its chapter.
No black boxes. IBM Model 1's EM loop, Bahdanau attention, the Transformer, BLEU — all implemented in plain Python you can read in one sitting.
Parts 0–2 need only the Python standard library. Every neural model trains in under 30 minutes on a free Colab T4 — or loads a pre-trained checkpoint instantly.
Each chapter ends by translating the same 20 held-out sentences. Seventy-five years of progress becomes something you can see, line by line.
Eight parts, in historical order. Each chapter is an executable notebook — open it here, in Colab, or in a Codespace.
The intellectual prehistory: information theory, the Weaver memorandum, and the first impossibility argument.
Dictionaries plus hand-written rules: the Georgetown demo, the Soviet mirror program, SYSTRAN, and interlingua.
Nagao's insight: don't translate from rules — translate by analogy to sentences you've seen before.
The IBM revolution: translation as probabilistic inference. EM, alignments, phrase tables, n-gram LMs.
Sequence-to-sequence learning, the fixed-vector bottleneck, and the attention mechanism that broke it.
Self-attention end to end: the architecture that ate NLP, plus the subword revolution and the scaling playbook.
One model, many languages: zero-shot transfer, M2M-100, NLLB-200, and translation as an emergent LLM skill.
How do we even know a translation is good? BLEU, chrF, TER and COMET implemented from scratch, plus human protocols.