tokenizer_config.json

{
  "do_lower_case": true,
  "strip_accents": false,
  "model_max_length": 2048,
  "tokenizer_class": "BertTokenizer"
}