tokenizer_config.json

{
  "vocab_size": 37,
  "pad_token": "<|pad|>",
  "eos_token": "<|endoftext|>",
  "unk_token": "<|unk|>",
  "bos_token": "<|endoftext|>"
}