PRIX / src /tokenizer.ts
Rachit-Tw's picture
Upload 169 files
9284ad7 verified
Raw
History Blame Contribute Delete
367 Bytes
// eslint-disable-next-line camelcase
import {get_encoding} from '@dqbd/tiktoken'
const tokenizer = get_encoding('cl100k_base')
export function encode(input: string): Uint32Array {
return tokenizer.encode(input)
}
export function getTokenCount(input: string): number {
input = input.replace(/<\|endoftext\|>/g, '')
return encode(input).length
}