Winoground

Overview edit

Variants: Winoground

Associated Benchmarks

This dataset is used in 1 benchmark:

• Visual Reasoning - Metrics: Text Score, Image Score, Group Score

Recent Benchmark Submissions

Research Papers

Recent papers with results on this dataset:

• A Cognitive Paradigm Approach to Probe the Perception-Reasoning Interface in VLMs (2025) -
• Prompting Large Vision-Language Models for Compositional Reasoning (2024) -
• CoCoT: Contrastive Chain-of-Thought Prompting for Large Multimodal Models with Multiple Image Inputs (2024) -
• Compositional Chain-of-Thought Prompting for Large Multimodal Models (2023) -
• SelfEval: Leveraging the discriminative nature of generative models for evaluation (2023) -
• The Role of Chain-of-Thought in Complex Vision-Language Reasoning Task (2023) -
• MMICL: Empowering Vision-language Model with Multi-Modal In-Context Learning (2023) -
• An Examination of the Compositionality of Large Generative Vision-Language Models (2023) -
• Revisiting the Role of Language Priors in Vision-Language Models (2023) -
• What You See is What You Read? Improving Text-Image Alignment Evaluation (2023) -
• Measuring Progress in Fine-grained Vision-and-Language Understanding (2023) -
• Simple Token-Level Confidence Improves Caption Correctness (2023) -
• Incorporating Structured Representations into Pretrained Vision & Language Models Using Scene Graphs (2023) -
• Going Beyond Nouns With Vision & Language Models Using Synthetic Data (2023) -
• Your Diffusion Model is Secretly a Zero-Shot Classifier (2023) -
• Equivariant Similarity for Vision-Language Foundation Models (2023) -
• Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality (2022) -