tokenizer/meta.json

{
  "original_vocab_size": 64000,
  "new_vocab_size": 64003,
  "special_tokens": {
    "<|user|>": 64000,
    "<|assistant|>": 64001,
    "<|endofturn|>": 64002
  }
}