tokenizer_config.json

{
  "ambiguous_token_ids": [
    1,
    6,
    7,
    8,
    9,
    10,
    11
  ],
  "auto_map": {
    "AutoTokenizer": [
      "tokenizer.ProteinTokenizer",
      null
    ]
  },
  "backend": "tokenizers",
  "bos_token": "<bos>",
  "bos_token_id": 3,
  "eos_token": "<eos>",
  "eos_token_id": 4,
  "mask_token": "<mask>",
  "mask_token_id": 2,
  "max_length": 2048,
  "model_input_names": [
    "input_ids",
    "attention_mask",
    "special_tokens_mask"
  ],
  "model_max_length": 2048,
  "pad_token": "<pad>",
  "pad_token_id": 0,
  "padding_side": "right",
  "tokenizer_class": "ProteinTokenizer",
  "truncation_side": "right",
  "unk_token": "<unk>",
  "unk_token_id": 1
}