README.md

metadata

pretty_name: OpenOrca
configs:
  - config_name: default
    default: true
    data_files:
      - split: train
        path:
          - '*.parquet'

OpenOrca but just the GPT4 bits.