tokenizer_config.json

{
  "add_prefix_space": false,
  "additional_special_tokens": [
    "[SYSTEM]",
    "[ASSISTANT]",
    "[USER]",
    "[END]"
  ],
  "bos_token": "<|endoftext|>",
  "clean_up_tokenization_spaces": true,
  "eos_token": "<|endoftext|>",
  "model_max_length": 1000,
  "pad_token": "[PAD]",
  "return_token_type_ids": false,
  "tokenizer_class": "GPT2Tokenizer",
  "unk_token": "<|endoftext|>",
  "vocab_size": 49152
}