– : –

Part-1 focused on how raw text becomes vectors the model can reason about — covering tokenization, subword units (BPE), and embedding vectors.

Part 2 looks at the next important piece of the pipeline: ?