tokenizer_config.json

{
  "bos_token": "^",
  "eos_token": "_",
  "model_max_length": 128,
  "pad_token": " ",
  "padding_side": "right",
  "special_tokens_map_file": "../../tokenizers/char_unigram_88/special_tokens_map.json",
  "tokenizer_class": "PreTrainedTokenizerFast",
  "truncation_side": "left",
  "unk_token": "§"
}