tokenizer_config.json

{
  "tokenizer_class": "PreTrainedTokenizerFast",
  "vocab_size": 32000,
  "model_max_length": 2048,
  "padding_side": "left",
  "truncation_side": "right",
  "special_tokens": {
    "<pad>": 0,
    "<bos>": 1,
    "<eos>": 2,
    "<unk>": 3
  },
  "bos_token": "<bos>",
  "eos_token": "<eos>",
  "pad_token": "<pad>",
  "unk_token": "<unk>"
}