Tokenization and the Embedding Table

Explain how a subword tokenizer splits text into token IDs and how an embedding lookup converts those IDs into a sequence of vectors; predict the shape of the resulting matrix for a given input.