tokenizer_config.json

{
    "bos_token": "<|begin_of_text|>",
    "clean_up_tokenization_spaces": false,
    "eos_token": "<|end_of_text|>",
    "extra_special_tokens": {},
    "model_input_names": [
      "input_ids",
      "attention_mask"
    ],
    "model_max_length": 131072,
    "pad_token": "<|end_of_text|>",
    "tokenizer_class": "PreTrainedTokenizerFast",
    "truncation_side": "left",
    "use_fast": true
  }