tokenizer_config.json

{
  "add_prefix_space": true,
  "additional_special_tokens": [
    "<e>",
    "</e>",
    "<a1>",
    "</a1>",
    "<a2>",
    "</a2>",
    "<cr>",
    "<neg>"
  ],
  "bos_token": "<s>",
  "cls_token": "<s>",
  "eos_token": "</s>",
  "errors": "replace",
  "mask_token": "<mask>",
  "model_max_length": 512,
  "name_or_path": "roberta-base",
  "pad_token": "<pad>",
  "sep_token": "</s>",
  "special_tokens_map_file": null,
  "tokenizer_class": "RobertaTokenizer",
  "trim_offsets": true,
  "unk_token": "<unk>"
}