import { get_encoding as getEncoding } from "@dqbd/tiktoken" const tokenizer = getEncoding("cl100k_base") const encode = (input: string): Uint32Array => { return tokenizer.encode(input) } const getTokenCount = (input: string): number => { const cleanedInput = input.replace(/<\|endoftext\|>/g, "") return encode(cleanedInput).length } const tokenTools = { getTokenCount, encode, } export default tokenTools