config.json

{
  "model_type": "sentencepiece",
  "tokenizer_class": "PreTrainedTokenizerFast",
  "vocab_size": 128000,
  "model_max_length": 512,
  "unk_token": "<unk>",
  "bos_token": "<s>",
  "eos_token": "</s>",
  "pad_token": "<pad>"
}