Implement byte-level BPE tokenization (minbpe)

Train byte-level BPE merges, encode and decode text, and inspect how the merges split rare words, numbers, code and non-English text.

The exercise happens on Andrej Karpathy (GitHub). Canonical link: https://github.com/karpathy/minbpe. Link check: link responded (5 Oct 2026). Opened from this app 0 times; opening is not counted as an attempt.

Supports: Tokenization questions in LLM interviews; a smaller warm-up for the CS336 tokenizer assignment.