Datasets:

Brendan
/

babylm-processed

dataset_info:
  features:
    - name: input_ids
      sequence: int32
    - name: attention_mask
      sequence: int8
    - name: special_tokens_mask
      sequence: int8
  splits:
    - name: train
      num_bytes: 6405468
      num_examples: 2077
    - name: valid
      num_bytes: 1612932
      num_examples: 523
  download_size: 2868688
  dataset_size: 8018400

Dataset Card for "babylm-processed"

More Information needed