---
dataset_info:
  features:
  - name: tokens
    sequence: int64
  - name: loss_mask
    sequence: float64
  - name: topic
    dtype: string
  splits:
  - name: train
    num_bytes: 559035571.7180814
    num_examples: 88276
  - name: validation
    num_bytes: 42081555.281918645
    num_examples: 6645
  download_size: 65557445
  dataset_size: 601117127.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
---