~/runtimelab $ cat ./text-generator/README
Build your own Text Generator
Build a text generator from counting alone — bigrams, sampling, temperature, top-k and top-p, then n-grams, smoothing, a repetition penalty and beam search. Every decoding knob in a modern LLM, with no neural network underneath.
what you end up with
A command-line text generator you can steer with temperature, top-k, top-p and a repetition penalty — and that stops on its own
- stages
- 15
- estimated
- 6–9 h
- level
- Beginner
- track
- AI Foundations
- language
- Python
- counting and conditional probability
- greedy vs sampled decoding
- temperature
- top-k and nucleus (top-p) sampling
- n-gram contexts
- add-k smoothing and backoff
- perplexity
- repetition penalties
- end-of-sequence tokens
- beam search
The stage list opens once you sign in — it is the challenge, not the pitch. Signing in is free, and the first stages of every challenge are too.