Interactive Scientific Simulator #charan

Inside the Transformer Forward Pass

Watch how the sentence “Charan planted a new thought in Advaitha’s head.” travels through every layer of a modern Transformer—from raw text characters to tokenization, embeddings, QKV attention, multi-head parallel lenses, residual connections, and next-token generation.

Speed

Amber = Focus & Query Purple = Values & Embeddings Teal = Keys & Positions Coral = Queries 💡 Click any token chip during attention stages to inspect its row

← Back to Transformer Overview 📝 Charan's Attention Notes 📝 Charan's Self-Attention Notes