tokenizer_config.json

{
  "tokenizer_class": "SentencePieceTokenizerWrapper",
  "vocab_size": 16000,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "pad_token_id": 0,
  "unk_token_id": 3,
  "model_max_length": 512
}