tokenizer_config.json

{
  "name_or_path": "THUDM/glm-roberta-large",
  "eos_token": "</s>",
  "pad_token": "</s>",
  "cls_token": "<s>",
  "mask_token": "[MASK]",
  "unk_token": "<unk>",
  "additional_special_tokens": ["<|startofpiece|>", "<|endofpiece|>"],
  "add_prefix_space": false,
  "tokenizer_class": "GLMRobertaTokenizer",
  "use_fast": false,
  "auto_map": {
    "AutoTokenizer": [
      "tokenization_glm.GLMRobertaTokenizer",
      null
      ]
  }
}