kdercksen's picture
Librarian Bot: Add language metadata for dataset (#2)
93c4742 verified
|
Raw
History Blame Contribute Delete
939 Bytes
metadata
language:
  - en
dataset_info:
  features:
    - name: sentence
      dtype: string
    - name: label
      dtype: string
  splits:
    - name: train
      num_bytes: 454998
      num_examples: 3560
    - name: validation
      num_bytes: 63674
      num_examples: 463
    - name: test
      num_bytes: 114554
      num_examples: 888
  download_size: 354318
  dataset_size: 633226
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*

This data is made available as part of the work Revise and Resubmit (https://doi.org/10.1162/coli_a_00455). Specifically, I have processed prag.csv from the project Github page (https://github.com/UKPLab/f1000rd) to be easily used for sentence classification using HuggingFace models. The sentences from the datasets are split into train/dev/test using the original splits.csv file from the Github repo.