Back to datasets
Dataset assetOpen Source CommunityCommon-sense ReasoningText Understanding

allenai/winogrande

WinoGrande is a dataset containing 44,000 problems, inspired by the Winograd Schema Challenge, but scaled up and adjusted to improve robustness against dataset‑specific biases. The task is cloze, providing two options; the goal is to select the correct option for the given sentence, requiring common‑sense reasoning.

Source
hugging_face
Created
Nov 28, 2025
Updated
Jan 18, 2024
Signals
459 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Overview

Dataset Name: WinoGrande

Dataset Configurations:

  • winogrande_xs
  • winogrande_s
  • winogrande_m
  • winogrande_l
  • winogrande_xl
  • winogrande_debiased

Dataset Features:

  • sentence: string
  • option1: string
  • option2: string
  • answer: string

Dataset Splits:

ConfigurationTrainValidationTest
winogrande_debiased924812671767
winogrande_l1023412671767
winogrande_m255812671767
winogrande_s64012671767
winogrande_xl4039812671767
winogrande_xs16012671767

Dataset Size:

  • Download size: 3,395,492 bytes
  • Dataset size varies by configuration, e.g.:
    • winogrande_xs: 412,552 bytes
    • winogrande_s: 474,156 bytes
    • winogrande_m: 720,849 bytes
    • winogrande_l: 1,711,424 bytes
    • winogrande_xl: 5,577,680 bytes
    • winogrande_debiased: 1,595,268 bytes

Citation:

@InProceedings{ai2:winogrande, title = {WinoGrande: An Adversarial Winograd Schema Challenge at Scale}, author = {Keisuke Sakaguchi and Ronan Le Bras and Chandra Bhagavatula and Yejin Choi}, year = {2019} }

Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio