tokenizer_config.json

{
  "num_layers": 9,
  "codebook_size": 1024,
  "vocab_size": 9221,
  "bos_token": "<|bos|>",
  "pad_token": "<|pad|>",
  "eos_token": "<|eos|>",
  "start_clean_token": "<|start_clean|>",
  "unk_token": "<|unk|>",
  "special_tokens": [
    "<|bos|>",
    "<|pad|>",
    "<|eos|>",
    "<|start_clean|>",
    "<|unk|>"
  ],
  "tokenizer_type": "DACTokenizer"
}