tokenizer_config.json

{
  "added_tokens_decoder": {
    "7886": {
      "content": "<s>",
      "lstrip": false,
      "normalized": false,
      "rstrip": false,
      "single_word": false,
      "special": true
    },
    "7887": {
      "content": "</s>",
      "lstrip": false,
      "normalized": false,
      "rstrip": false,
      "single_word": false,
      "special": true
    },
    "7888": {
      "content": "<unk>",
      "lstrip": false,
      "normalized": false,
      "rstrip": false,
      "single_word": false,
      "special": true
    },
    "7889": {
      "content": "<pad>",
      "lstrip": false,
      "normalized": false,
      "rstrip": false,
      "single_word": false,
      "special": true
    }
  },
  "auto_map": {
    "AutoTokenizer": [
      "tokenmonster_hf.TokenMonsterTokenizer",
      null
    ]
  },
  "backend": "custom",
  "bos_token": "<s>",
  "eos_token": "</s>",
  "model_max_length": 1000000,
  "pad_token": "<pad>",
  "tokenizer_class": "TokenMonsterTokenizer",
  "unk_token": "<unk>",
  "vocab_file": "tokenmonster.vocab"
}