tokenizer_config.json

{
  "name_or_path": "THUDM/glm-large-chinese",
  "eos_token": "<|endoftext|>",
  "pad_token": "<|endoftext|>",
  "cls_token": "[CLS]",
  "mask_token": "[MASK]",
  "unk_token": "[UNK]",
  "additional_special_tokens": ["<|startofpiece|>", "<|endofpiece|>", "[gMASK]", "[sMASK]"],
  "add_prefix_space": false,
  "tokenizer_class": "GLMChineseTokenizer",
  "use_fast": false,
  "auto_map": {
    "AutoTokenizer": [
      "tokenization_glm.GLMChineseTokenizer",
      null
      ]
  }
}