---
dataset_info:
- config_name: dart
  features:
  - name: gem_id
    dtype: string
  - name: gem_parent_id
    dtype: string
  - name: dart_id
    dtype: int32
  - name: tripleset
    list:
      list: string
  - name: subtree_was_extended
    dtype: bool
  - name: target_sources
    list: string
  - name: target
    dtype: string
  - name: references
    list: string
  splits:
  - name: test
    num_bytes: 3477176
    num_examples: 5097
  - name: train
    num_bytes: 23051538
    num_examples: 62659
  - name: validation
    num_bytes: 1934235
    num_examples: 2768
  download_size: 19746789
  dataset_size: 28462949
- config_name: e2e_nlg
  features:
  - name: gem_id
    dtype: string
  - name: gem_parent_id
    dtype: string
  - name: meaning_representation
    dtype: string
  - name: target
    dtype: string
  - name: references
    list: string
  splits:
  - name: challenge_test_scramble
    num_bytes: 236175
    num_examples: 500
  - name: challenge_train_sample
    num_bytes: 145295
    num_examples: 500
  - name: challenge_validation_sample
    num_bytes: 226501
    num_examples: 500
  - name: test
    num_bytes: 2133671
    num_examples: 4693
  - name: train
    num_bytes: 9128934
    num_examples: 33525
  - name: validation
    num_bytes: 1856073
    num_examples: 4299
  download_size: 4409576
  dataset_size: 13726649
- config_name: web_nlg_en
  features:
  - name: gem_id
    dtype: string
  - name: gem_parent_id
    dtype: string
  - name: input
    list: string
  - name: target
    dtype: string
  - name: references
    list: string
  - name: category
    dtype: string
  - name: webnlg_id
    dtype: string
  splits:
  - name: challenge_test_numbers
    num_bytes: 409177
    num_examples: 500
  - name: challenge_test_scramble
    num_bytes: 402371
    num_examples: 500
  - name: challenge_train_sample
    num_bytes: 193162
    num_examples: 502
  - name: challenge_validation_sample
    num_bytes: 359832
    num_examples: 499
  - name: test
    num_bytes: 1403565
    num_examples: 1779
  - name: train
    num_bytes: 13067471
    num_examples: 35426
  - name: validation
    num_bytes: 1153959
    num_examples: 1667
  download_size: 10366803
  dataset_size: 16989537
configs:
- config_name: dart
  data_files:
  - split: test
    path: dart/test-*
  - split: train
    path: dart/train-*
  - split: validation
    path: dart/validation-*
- config_name: e2e_nlg
  data_files:
  - split: challenge_test_scramble
    path: e2e_nlg/challenge_test_scramble-*
  - split: challenge_train_sample
    path: e2e_nlg/challenge_train_sample-*
  - split: challenge_validation_sample
    path: e2e_nlg/challenge_validation_sample-*
  - split: test
    path: e2e_nlg/test-*
  - split: train
    path: e2e_nlg/train-*
  - split: validation
    path: e2e_nlg/validation-*
- config_name: web_nlg_en
  data_files:
  - split: challenge_test_numbers
    path: web_nlg_en/challenge_test_numbers-*
  - split: challenge_test_scramble
    path: web_nlg_en/challenge_test_scramble-*
  - split: challenge_train_sample
    path: web_nlg_en/challenge_train_sample-*
  - split: challenge_validation_sample
    path: web_nlg_en/challenge_validation_sample-*
  - split: test
    path: web_nlg_en/test-*
  - split: train
    path: web_nlg_en/train-*
  - split: validation
    path: web_nlg_en/validation-*
license: other
---

# gem

A copy of [GEM/gem](https://huggingface.co/datasets/GEM/gem) in Parquet format. The original uses a dataset loading script, which `datasets>=4` no longer supports. The data is unchanged: it was exported with `datasets==3.6.0` using the original script.

Please cite and follow the license of the original dataset. Subsets keep their original licenses: `web_nlg_en` (WebNLG) is CC BY-NC-SA 4.0, `dart` is MIT, `e2e_nlg` is CC BY-SA 4.0.