ML Research Wiki / Benchmarks / Explanation Generation / WHOOPS!

WHOOPS!

Explanation Generation Benchmark

Performance Over Time

đź“Š Showing 7 results | đź“Ź Metric: Human (%)

Top Performing Models

Rank Model Paper Human (%) Date Code
1 VLIS (Lynx) VLIS: Unimodal Language Models Guide Multimodal Language Generation 80.00 2023-10-15 📦 jiwanchung/vlis
2 VLIS (LLaVA) VLIS: Unimodal Language Models Guide Multimodal Language Generation 73.00 2023-10-15 📦 jiwanchung/vlis
3 Ground-truth Caption -> GPT3 (Oracle) Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 68.00 2023-03-13 -
4 Predicted Caption -> GPT3 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 33.00 2023-03-13 -
5 BLIP2 FlanT5-XXL (Fine-tuned) đź“š Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 27.00 2023-03-13 -
6 BLIP2 FlanT5-XL (Fine-tuned) đź“š Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 15.00 2023-03-13 -
7 BLIP2 FlanT5-XXL (Zero-shot) Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 0.00 2023-03-13 -

All Papers (7)

jiwanchung/vlis
jiwanchung/vlis