Encoder
Domain: LLM Concepts
🎯 Motivation
The encoder part of a transformer processes the entire input simultaneously, building a rich understanding (contextualized representations) of the text.
📋 Concrete Example
Input ‘The cat sat’ goes through encoder layers. Each token’s representation is enriched with context from other tokens. Output: deep understanding vectors for each token.
🔗 Analogy
A scholar studying a text - they read the whole thing, cross-reference, take notes, and build a deep understanding. The encoder does this mathematical version of deep reading.