---
dataset_info:
  features:
  - name: text
    dtype: string
  - name: meta
    struct:
    - name: pile_set_name
      dtype: string
  splits:
  - name: train
    num_bytes: 3221975934
    num_examples: 100000
  download_size: 1488397105
  dataset_size: 3221975934
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---


## Dataset Creation Process
These subsets were created by streaming over the rows from [`monology/pile-uncopyrighted`](https://huggingface.co/datasets/monology/pile-uncopyrighted) and filtering by the meta column. Each subset is generally limited to the first 100,000 qualifying rows encountered.