tokenizer_config.json

{
  "tokenizer_class": "PreTrainedTokenizerFast",
  "auto_map": {
    "AutoTokenizer": [
      "tokenizers.Tokenizer",
      null
    ]
  },
  "model_type": "bpe",
  "vocab_size": 32000,
  "unk_token": "[UNK]",
  "special_tokens": {
    "unk_token": "[UNK]",
    "additional_special_tokens": [
      "<NUM>",
      "<URL>",
      "<EMAIL>"
    ]
  },
  "model_max_length": 512,
  "padding_side": "right",
  "truncation_side": "right",
  "clean_up_tokenization_spaces": true
}