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

Visual Reasoning on NLVR2 v2 (dev)

88.7Accuracy

X2-VLM_large

37.42850.73964.0577.361Nov 22, 2022Jan 26, 2023Apr 1, 2023Jun 5, 2023Aug 9, 2023Oct 13, 2023Dec 18, 2023
Updated 1mo ago

Evaluation Results

MethodLinks
2022.11
88.7
2022.11
87.2
2022.11
86.2
2022.11
85.9
2022.11
85.6
2022.11
84.1
2022.11
82.8
2022.11
82.5
2022.11
82.3
2022.11
81.9
2022.11
81.7
2022.11
80.2
2023.12
65.6
2023.12
62.2
2023.12
60.8
2023.12
51.6
2023.12
48.2
2023.12
45.4
2023.12
44
2023.12
39.4