metadata

dataset_info:
  features:
    - name: input_ids
      sequence: int32
    - name: token_type_ids
      sequence: int8
    - name: attention_mask
      sequence: int8
    - name: labels
      sequence: int64
  splits:
    - name: train
      num_bytes: 144640000
      num_examples: 80000
    - name: validation
      num_bytes: 14464000
      num_examples: 8000
    - name: test
      num_bytes: 1446400
      num_examples: 800
  download_size: 24687721
  dataset_size: 160550400

Dataset Card for "BertTokenizer_THUCNews_10000_to_lm_datasets"

选取seamew/THUCNewsText数据集10000条，微调IDEA-CCNL/Wenzhong2.0-GPT2-110M-BertTokenizer-chinese模型

More Information needed