Datasets:

Modalities:
Text
Formats:
parquet
ArXiv:
Libraries:
Datasets
pandas
ViNLI_remake / README.md
npvinHnivqn's picture
Update README.md
ad8430b verified
metadata
dataset_info:
  features:
    - name: entailment
      sequence: string
    - name: contradiction
      sequence: string
    - name: other
      sequence: string
    - name: neutral
      sequence: string
    - name: anchor
      dtype: string
  splits:
    - name: train
      num_bytes: 3832846
      num_examples: 3036
    - name: test
      num_bytes: 446604
      num_examples: 363
    - name: dev
      num_bytes: 449297
      num_examples: 367
  download_size: 2794159
  dataset_size: 4728747
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
      - split: dev
        path: data/dev-*

This dataset is created to evaluate language models on their retrieval and re-ranking abilities. More detailed information can be found at our publication: Advancing Vietnamese Information Retrieval with Learning Objective and Benchmark.

Please cite this paper if you want to use it for your work.