tokenizer_config.json

{
  "clean_up_tokenization_spaces": true,
  "model_max_length": 2048,
  "padding_side": "right",
  "tokenizer_class": "PreTrainedTokenizerFast"
}