tokenizer_config.json

{
  "bos_token": "<|endoftext|>",
  "eos_token": "<|endoftext|>",
  "model_max_length": 1000000000000000019884624838656,
  "tokenizer_class": "PreTrainedTokenizerFast"
}