README.md

metadata

language:
  - en
license: odc-by
dataset_info:
  features:
    - name: text
      dtype: string
    - name: id
      dtype: string
    - name: dump
      dtype: string
    - name: url
      dtype: string
    - name: file_path
      dtype: string
    - name: language
      dtype: string
    - name: language_score
      dtype: float64
    - name: token_count
      dtype: int64
    - name: score
      dtype: float64
    - name: int_score
      dtype: int64
  splits:
    - name: train
      num_bytes: 2003951.6881090973
      num_examples: 394
  download_size: 2733459
  dataset_size: 2003951.6881090973
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

FineWeb-Edu Micro

This dataset is a subset of the FineWeb-Edu Sample-10BT, which contains passages that are at least 1000 tokens long, totalling about 1 Million tokens .

This dataset was primarily made to evaluate different RAG Chunking mechanisms in Chonkie