{
  "format": "m5-caption-accent-speaker-lang-v3",
  "field_order": [
    "caption",
    "accent",
    "speaker",
    "lang",
    "text"
  ],
  "missing_condition": {
    "caption": "empty_between_delimiters",
    "accent": "empty_between_delimiters",
    "speaker": "literal_None",
    "lang": "empty_between_delimiters"
  },
  "base_vocab_size": 129984,
  "vocab_size": 129992,
  "control_token_ids": {
    "<|caption_begin|>": 129984,
    "<|caption_end|>": 129985,
    "<|lang_begin|>": 129986,
    "<|lang_end|>": 129987,
    "<|accent_begin|>": 129988,
    "<|accent_end|>": 129989,
    "<|speaker_start|>": 129990,
    "<|speaker_end|>": 129991
  },
  "vocab_sha256": "05fd46dd14d5234c102aba957794a1e3a308fc86dc1c96be13db71cec4a5dbfe",
  "tokenizer_pipeline_sha256": "a0a37555c0450564e132336a43e26b0989900268b72fb74eb64a9a079dd91fbc",
  "bos_token_id": 128000,
  "eos_token_id": 128001
}