README.md

metadata

dataset_info:
  features:
    - name: input_ids
      sequence: int32
    - name: attention_mask
      sequence: int8
    - name: labels
      sequence: int64
  splits:
    - name: train
      num_bytes: 53397189200
      num_examples: 2004700
    - name: test
      num_bytes: 532720000
      num_examples: 20000
  download_size: 16064350610
  dataset_size: 53929909200
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*