tokenizer_config.json

{
  "tokenizer_class": "LlamaTokenizer",
  "vocab_size": 50000,
  "model_max_length": 2048,
  "padding_side": "right",
  "truncation_side": "right",
  "use_fast": false,
  "clean_up_tokenization_spaces": true,
  "bos_token": "<s>",
  "eos_token": "</s>",
  "unk_token": "<unk>",
  "pad_token": "<pad>",
  "mask_token": "<mask>",
  "cls_token": "<cls>",
  "sep_token": "</s>",
  "additional_special_tokens": [
    "<s>",
    "</s>",
    "<unk>",
    "<pad>"
  ],
  "special_tokens_map": {
    "<unk>": "<unk>",
    "<s>": "<s>",
    "</s>": "</s>",
    "<pad>": "<pad>"
  }
}