tokenizer_config.json

{
  "clean_up_tokenization_spaces": true,
  "cls_token": "<START>",
  "do_lower_case": true,
  "mask_token": "[MASK]",
  "model_input_names": [
    "input_ids",
    "attention_mask",
    "decoder_input_ids"
  ],
  "model_max_length": 1024,
  "pad_token": "<PAD>",
  "sep_token": "<END>",
  "strip_accents": null,
  "tokenize_chinese_chars": true,
  "tokenizer_class": "BertTokenizer",
  "unk_token": "<UNK>"
}