BPE
Domain: NLP Fundamentals
🎯 Motivation
Byte-Pair Encoding starts with characters and merges the most frequent pairs step by step. Used by GPT models.
📋 Concrete Example
Start with characters. low + est merge to lowest if seen often. Build a vocabulary of frequent subwords.
🔗 Analogy
Like a factory that starts with tiny parts and keeps gluing the most common pairs together until it has useful modules.