babi_nli / README.md
sileod's picture
Upload README.md with huggingface_hub
8e567c6 verified
|
raw
history blame
13.2 kB
metadata
annotations_creators:
  - expert-generated
language_creators:
  - crowdsourced
language:
  - en
license: bsd
multilinguality:
  - monolingual
size_categories:
  - 1K<n<10K
source_datasets:
  - original
task_categories:
  - text-classification
task_ids:
  - natural-language-inference
pretty_name: babi_nli
tags:
  - logical reasoning
  - nli
  - natural-language-inference
  - reasoning
  - logic
dataset_info:
  - config_name: basic-coreference
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 247359
        num_examples: 1000
      - name: validation
        num_bytes: 125770
        num_examples: 500
      - name: test
        num_bytes: 122865
        num_examples: 500
    download_size: 98403
    dataset_size: 495994
  - config_name: basic-deduction
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 225519
        num_examples: 1000
      - name: validation
        num_bytes: 112981
        num_examples: 500
      - name: test
        num_bytes: 112862
        num_examples: 500
    download_size: 49023
    dataset_size: 451362
  - config_name: basic-induction
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 188319
        num_examples: 1000
      - name: validation
        num_bytes: 94081
        num_examples: 500
      - name: test
        num_bytes: 94090
        num_examples: 500
    download_size: 85153
    dataset_size: 376490
  - config_name: compound-coreference
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 282687
        num_examples: 1000
      - name: validation
        num_bytes: 139161
        num_examples: 500
      - name: test
        num_bytes: 143924
        num_examples: 500
    download_size: 109694
    dataset_size: 565772
  - config_name: conjunction
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 284487
        num_examples: 1000
      - name: validation
        num_bytes: 141534
        num_examples: 500
      - name: test
        num_bytes: 142683
        num_examples: 500
    download_size: 118704
    dataset_size: 568704
  - config_name: counting
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 325802
        num_examples: 1000
      - name: validation
        num_bytes: 154693
        num_examples: 500
      - name: test
        num_bytes: 154519
        num_examples: 500
    download_size: 119022
    dataset_size: 635014
  - config_name: indefinite-knowledge
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 227391
        num_examples: 1000
      - name: validation
        num_bytes: 113140
        num_examples: 500
      - name: test
        num_bytes: 113298
        num_examples: 500
    download_size: 96252
    dataset_size: 453829
  - config_name: lists-sets
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 301224
        num_examples: 1000
      - name: validation
        num_bytes: 149567
        num_examples: 500
      - name: test
        num_bytes: 153024
        num_examples: 500
    download_size: 119319
    dataset_size: 603815
  - config_name: positional-reasoning
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 158135
        num_examples: 1000
      - name: validation
        num_bytes: 79034
        num_examples: 500
      - name: test
        num_bytes: 78998
        num_examples: 500
    download_size: 34376
    dataset_size: 316167
  - config_name: simple-negation
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 218297
        num_examples: 1000
      - name: validation
        num_bytes: 110967
        num_examples: 500
      - name: test
        num_bytes: 107212
        num_examples: 500
    download_size: 88932
    dataset_size: 436476
  - config_name: single-supporting-fact
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 223761
        num_examples: 1000
      - name: validation
        num_bytes: 112784
        num_examples: 500
      - name: test
        num_bytes: 111569
        num_examples: 500
    download_size: 91968
    dataset_size: 448114
  - config_name: three-arg-relations
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 598485
        num_examples: 1000
      - name: validation
        num_bytes: 312900
        num_examples: 500
      - name: test
        num_bytes: 324225
        num_examples: 500
    download_size: 243096
    dataset_size: 1235610
  - config_name: three-supporting-facts
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 1479097
        num_examples: 1000
      - name: validation
        num_bytes: 783450
        num_examples: 500
      - name: test
        num_bytes: 735719
        num_examples: 500
    download_size: 558073
    dataset_size: 2998266
  - config_name: time-reasoning
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 379567
        num_examples: 1000
      - name: validation
        num_bytes: 188913
        num_examples: 500
      - name: test
        num_bytes: 187157
        num_examples: 500
    download_size: 152925
    dataset_size: 755637
  - config_name: two-arg-relations
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 122565
        num_examples: 1000
      - name: validation
        num_bytes: 61228
        num_examples: 500
      - name: test
        num_bytes: 61170
        num_examples: 500
    download_size: 54533
    dataset_size: 244963
  - config_name: two-supporting-facts
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 507085
        num_examples: 1000
      - name: validation
        num_bytes: 261742
        num_examples: 500
      - name: test
        num_bytes: 244322
        num_examples: 500
    download_size: 194986
    dataset_size: 1013149
  - config_name: yes-no-questions
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype:
          class_label:
            names:
              '0': not-entailed
              '1': entailed
      - name: idx
        dtype: int32
    splits:
      - name: train
        num_bytes: 230579
        num_examples: 1000
      - name: validation
        num_bytes: 115915
        num_examples: 500
      - name: test
        num_bytes: 113022
        num_examples: 500
    download_size: 94399
    dataset_size: 459516
configs:
  - config_name: basic-coreference
    data_files:
      - split: train
        path: basic-coreference/train-*
      - split: validation
        path: basic-coreference/validation-*
      - split: test
        path: basic-coreference/test-*
  - config_name: basic-deduction
    data_files:
      - split: train
        path: basic-deduction/train-*
      - split: validation
        path: basic-deduction/validation-*
      - split: test
        path: basic-deduction/test-*
  - config_name: basic-induction
    data_files:
      - split: train
        path: basic-induction/train-*
      - split: validation
        path: basic-induction/validation-*
      - split: test
        path: basic-induction/test-*
  - config_name: compound-coreference
    data_files:
      - split: train
        path: compound-coreference/train-*
      - split: validation
        path: compound-coreference/validation-*
      - split: test
        path: compound-coreference/test-*
  - config_name: conjunction
    data_files:
      - split: train
        path: conjunction/train-*
      - split: validation
        path: conjunction/validation-*
      - split: test
        path: conjunction/test-*
  - config_name: counting
    data_files:
      - split: train
        path: counting/train-*
      - split: validation
        path: counting/validation-*
      - split: test
        path: counting/test-*
  - config_name: indefinite-knowledge
    data_files:
      - split: train
        path: indefinite-knowledge/train-*
      - split: validation
        path: indefinite-knowledge/validation-*
      - split: test
        path: indefinite-knowledge/test-*
  - config_name: lists-sets
    data_files:
      - split: train
        path: lists-sets/train-*
      - split: validation
        path: lists-sets/validation-*
      - split: test
        path: lists-sets/test-*
  - config_name: positional-reasoning
    data_files:
      - split: train
        path: positional-reasoning/train-*
      - split: validation
        path: positional-reasoning/validation-*
      - split: test
        path: positional-reasoning/test-*
  - config_name: simple-negation
    data_files:
      - split: train
        path: simple-negation/train-*
      - split: validation
        path: simple-negation/validation-*
      - split: test
        path: simple-negation/test-*
  - config_name: single-supporting-fact
    data_files:
      - split: train
        path: single-supporting-fact/train-*
      - split: validation
        path: single-supporting-fact/validation-*
      - split: test
        path: single-supporting-fact/test-*
  - config_name: three-arg-relations
    data_files:
      - split: train
        path: three-arg-relations/train-*
      - split: validation
        path: three-arg-relations/validation-*
      - split: test
        path: three-arg-relations/test-*
  - config_name: three-supporting-facts
    data_files:
      - split: train
        path: three-supporting-facts/train-*
      - split: validation
        path: three-supporting-facts/validation-*
      - split: test
        path: three-supporting-facts/test-*
  - config_name: time-reasoning
    data_files:
      - split: train
        path: time-reasoning/train-*
      - split: validation
        path: time-reasoning/validation-*
      - split: test
        path: time-reasoning/test-*
  - config_name: two-arg-relations
    data_files:
      - split: train
        path: two-arg-relations/train-*
      - split: validation
        path: two-arg-relations/validation-*
      - split: test
        path: two-arg-relations/test-*
  - config_name: two-supporting-facts
    data_files:
      - split: train
        path: two-supporting-facts/train-*
      - split: validation
        path: two-supporting-facts/validation-*
      - split: test
        path: two-supporting-facts/test-*
  - config_name: yes-no-questions
    data_files:
      - split: train
        path: yes-no-questions/train-*
      - split: validation
        path: yes-no-questions/validation-*
      - split: test
        path: yes-no-questions/test-*

bAbi_nli

bAbI tasks recasted as natural language inference. https://github.com/facebookarchive/bAbI-tasks

tasksource recasting code: https://colab.research.google.com/drive/1J_RqDSw9iPxJSBvCJu-VRbjXnrEjKVvr?usp=sharing

@article{weston2015towards,
  title={Towards ai-complete question answering: A set of prerequisite toy tasks},
  author={Weston, Jason and Bordes, Antoine and Chopra, Sumit and Rush, Alexander M and Van Merri{\"e}nboer, Bart and Joulin, Armand and Mikolov, Tomas},
  journal={arXiv preprint arXiv:1502.05698},
  year={2015}
}