This is only a tokenizer. This tokenizer is a GPT2TokenizerFast which is trained on raygx/Nepali-Extended-Corpus datasets. Code is available here.