paper-with-me

Visual Question Answering (VQA) 벤치마크

Visual Question Answering (VQA) on WHOOPS!

6개 결과 · ⬇ CSV · JSON

Exact Match

4 8.25 12.5 16.75 21 2023-03 2026-09 BLIP2 FlanT5-XXL (Fine-tuned) — 21.0 (2023-03-13) BLIP2 FlanT5-XL (Fine-tuned) — 20.0 (2023-03-13) BLIP2 FlanT5-XXL (Zero-shot) — 15.0 (2023-03-13) OFA Large — 8.0 (2023-03-13) BLIP Large — 6.0 (2023-03-13) BLIP2 FlanT5-XXL (Text-only FT) — 4.0 (2023-03-13) BLIP2 FlanT5-XXL (Fine-tuned) — 21.0 (2023-03-13)
RankModel Exact MatchBEM Extra Training Data PaperCodeYear
1 BLIP2 FlanT5-XXL (Fine-tuned) 2157 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 2023
2 BLIP2 FlanT5-XL (Fine-tuned) 2055 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 2023
3 BLIP2 FlanT5-XXL (Zero-shot) 1555 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 2023
4 OFA Large 838 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 2023
5 BLIP Large 639 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 2023
6 BLIP2 FlanT5-XXL (Text-only FT) 424 Breaking Common Sense: WHOOPS! A Vision-and-Language Benchmark of Synthetic and Compositional Images 2023
1–6 / 6 페이지당 10 20 50 100