slconnin's picture
V2.5 synthetic dataset with balanced diversity - expanded vocabulary, multiple text formats, reduced template predictability while maintaining clear category boundaries
f72f308 verified
metadata
dataset_info:
  features:
    - name: text
      dtype: string
    - name: label
      dtype: int64
    - name: label_name
      dtype: string
  splits:
    - name: train
      num_bytes: 1376576
      num_examples: 8000
    - name: validation
      num_bytes: 342839
      num_examples: 2000
    - name: test
      num_bytes: 340389
      num_examples: 2000
  download_size: 585135
  dataset_size: 2059804
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*