tokenizer_config.json

{
  "vocab_size": 32000,
  "model_max_length": 2048,
  "tokenizer_class": "BPETokenizer",
  "special_tokens": {
    "pad_token": "<|pad|>",
    "unk_token": "<|unk|>",
    "bos_token": "<|begin_of_sentence|>",
    "eos_token": "<|end_of_sentence|>"
  }
}