BLIP2 FlanT5-XXL (Text-only FT)

Reported on 3 benchmarks across 2 tasks · 1 paper · 1 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing3 results

Image-to-Text RetrievalonWHOOPS!
Specificity· uses extra data· 2023-03-13
94
SOTA
Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images arXiv:2303.07274
Visual Question Answering (VQA)onWHOOPS!
BEM· uses extra data· 2023-03-13
24
best: 57 (BLIP2 FlanT5-XXL (Fine-tuned))
Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images arXiv:2303.07274
Visual Question Answering (VQA)onWHOOPS!
Exact Match· uses extra data· 2023-03-13
4
best: 21 (BLIP2 FlanT5-XXL (Fine-tuned))
Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images arXiv:2303.07274