tokenizer_config.json

{
  "tokenizer_class": "LISATokenizer",
  "auto_map": {
    "AutoTokenizer": [
      "tokenizer.LISATokenizer",
      null
    ]
  },
  "bos_token": "<SOS>",
  "eos_token": "<EOS>",
  "unk_token": "<UNK>",
  "pad_token": "<PAD>",
  "model_max_length": 512,
  "special_tokens_map_file": "special_tokens_map.json",
  "vocab_file": "vocab.txt",
  "multimodal_support": true,
  "vision_token": "<IMG>",
  "audio_token": "<AUD>",
  "video_token": "<VID>",
  "lisa_metadata": {
    "developed_in": "Kenya, Africa",
    "team": "LISA Team"
  }
}