MicroGPT.ts - a conversion of Karpathy's MicroGPT to Typescript
    Preparing search index...

    Function buildTokenizer

    • Build a character-level tokenizer from a list of documents.

      Parameters

      • docs: string[]

        Array of document strings.

      Returns { BOS: number; uchars: string[]; vocabSize: number }

      uchars (sorted unique characters), BOS token id, and vocabSize.