indicxnli_repaired / README.md
SultanR's picture
Upload README.md with huggingface_hub
9ef25f8 verified
|
Raw
History Blame Contribute Delete
7.59 kB
metadata
dataset_info:
  - config_name: as
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: bn
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: gu
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: hi
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: kn
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: ml
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: mr
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: or
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: pa
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
  - config_name: ta
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 3238
      - name: test
        num_examples: 5010
  - config_name: te
    features:
      - name: premise
        dtype: string
      - name: hypothesis
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_examples: 392702
      - name: validation
        num_examples: 2490
      - name: test
        num_examples: 5010
configs:
  - config_name: as
    data_files:
      - split: train
        path: data/as/train-*
      - split: validation
        path: data/as/validation-*
      - split: test
        path: data/as/test-*
  - config_name: bn
    data_files:
      - split: train
        path: data/bn/train-*
      - split: validation
        path: data/bn/validation-*
      - split: test
        path: data/bn/test-*
  - config_name: gu
    data_files:
      - split: train
        path: data/gu/train-*
      - split: validation
        path: data/gu/validation-*
      - split: test
        path: data/gu/test-*
  - config_name: hi
    data_files:
      - split: train
        path: data/hi/train-*
      - split: validation
        path: data/hi/validation-*
      - split: test
        path: data/hi/test-*
  - config_name: kn
    data_files:
      - split: train
        path: data/kn/train-*
      - split: validation
        path: data/kn/validation-*
      - split: test
        path: data/kn/test-*
  - config_name: ml
    data_files:
      - split: train
        path: data/ml/train-*
      - split: validation
        path: data/ml/validation-*
      - split: test
        path: data/ml/test-*
  - config_name: mr
    data_files:
      - split: train
        path: data/mr/train-*
      - split: validation
        path: data/mr/validation-*
      - split: test
        path: data/mr/test-*
  - config_name: or
    data_files:
      - split: train
        path: data/or/train-*
      - split: validation
        path: data/or/validation-*
      - split: test
        path: data/or/test-*
  - config_name: pa
    data_files:
      - split: train
        path: data/pa/train-*
      - split: validation
        path: data/pa/validation-*
      - split: test
        path: data/pa/test-*
  - config_name: ta
    data_files:
      - split: train
        path: data/ta/train-*
      - split: validation
        path: data/ta/validation-*
      - split: test
        path: data/ta/test-*
  - config_name: te
    data_files:
      - split: train
        path: data/te/train-*
      - split: validation
        path: data/te/validation-*
      - split: test
        path: data/te/test-*

IndicXNLI (Repaired)

This is a repaired version of the Divyanshu/indicxnli dataset, converted to parquet format for compatibility with HuggingFace datasets 4.x+.

Why this exists

The original dataset uses a Python loading script (indicxnli.py) which is no longer supported in HuggingFace datasets 4.x. This version converts the data to native parquet format.

Original Dataset

Languages

Code Language
as Assamese
bn Bengali
gu Gujarati
hi Hindi
kn Kannada
ml Malayalam
mr Marathi
or Oriya
pa Punjabi
ta Tamil
te Telugu

Usage

from datasets import load_dataset

# Load Hindi validation split
ds = load_dataset("AdaMLLab/indicxnli_repaired", "hi", split="validation")

Schema

Each sample contains:

  • premise (string): The premise sentence
  • hypothesis (string): The hypothesis sentence
  • label (int): 0=entailment, 1=neutral, 2=contradiction

Data Source

This dataset uses the "forward" translation direction from the original dataset, where English XNLI was translated to Indic languages.

Citation

If you use this dataset, please cite the original paper:

@inproceedings{aggarwal-etal-2022-indicxnli,
    title = "{I}ndic{XNLI}: Evaluating Multilingual Inference for {I}ndian Languages",
    author = "Aggarwal, Divyanshu and Gupta, Vivek and Kunchukuttan, Anoop",
    booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
    year = "2022"
}