How LLMs Actually Think

Tokens, Embeddings, and the Magic of Transformers