tokenizer_config.json

{
  "add_prefix_space": false,
  "additional_special_tokens": [
    "<|system|>",
    "<|assistant|>",
    "<|user|>",
    "<|end|>",
    "<|PAD|>"
  ],
  "bos_token": "<|endoftext|>",
  "clean_up_tokenization_spaces": true,
  "eos_token": "<|endoftext|>",
  "model_max_length": 1400,
  "tokenizer_class": "CodeGenTokenizer",
  "unk_token": "<|endoftext|>"
}