Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

InfoSeek

Benchmarks

Task NameDataset NameSOTA ResultTrend
Visual Question AnsweringInfoSeek (test)
Accuracy57.9
81
Visual Question AnsweringInfoSeek
Accuracy69
77
Visual Question AnsweringInfoSeek (Full)
Accuracy43
61
Knowledge-Intensive Visual Question AnsweringInfoSeek (val)
Accuracy (All)44.1
50
Visual Question AnsweringInfoSeek
Unseen-Q Score47.8
49
Visual Question AnsweringInfoSeek (val)
Overall Accuracy47.2
45
Uncertainty QuantificationInfoSeek (val)
Accuracy14.7
42
Knowledge-based Visual Question AnsweringINFOSEEK Unseen Question
Accuracy46.5
42
Visual Information SeekingInfoSeek
Pass@174.8
24
Multi-hop information-seekingInfoSeek-Eval ID
Success Rate (SR)82
24
Visual Question AnsweringINFOSEEK Unseen-E
Accuracy42.1
23
Visual Question AnsweringInfoSeek
F1 Recall43.6
22
(Image, Text)-to-Text RetrievalInfoSeek
Recall@570.3
20
Knowledge-based Visual Question AnsweringINFOSEEK (Unseen Entity)
Accuracy51
19
Knowledge-Based Visual Question AnsweringINFOSEEK
FR (All)42.5
18
Knowledge-based VQAInfoSeek
Unseen-Q Performance42.49
18
Information Seeking Question AnsweringInfoSeek
Accuracy73.9
17
Knowledge-Based Visual Question AnsweringInfoSeek All
Accuracy49.9
16
RetrievalInfoSeek
Recall@159.6
12
Re-rankingInfoSeek
R@166.5
11
RetrievalInfoSeek standard (val)
Recall@167
10
Entity RetrievalInfoSeek (val)
R@164
9
Visual Question AnsweringInfoseek M2KR setup (test)
VQA Accuracy42.8
8
Visual Question AnsweringInfoSeek
Accuracy34.6
8
(Image, Text)-to-Multimodal RetrievalInfoSeek
Recall@548.9
8
Showing 25 of 41 rows