reasoning-19
Visual Entailment
3260 papers • 126 benchmarks • 313 datasets
Visual Entailment (VE) - is a task consisting of image-sentence pairs whereby a premise is defined by an image, rather than a natural language sentence as in traditional Textual Entailment tasks. The goal is to predict whether the image semantically entails the text.
(Image credit: Open Source)
Benchmarks
These leaderboards are used to track progress in visual-entailment-39
Trend
Dataset
Best Model
Actions
SNLI-VE val
SNLI-VE val
SNLI-VE test
SNLI-VE test
e-SNLI-VE
e-SNLI-VE
Libraries
iUse these libraries to find visual-entailment-39 models and implementations
2 papers 2,320
Datasets
No datasets available.
Subtasks
No subtasks available.