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

Natural Language Visual Reasoning on NLVR2

88.8Accuracy

GPT-4V

5.49627.12348.7570.377Sep 14, 2023Feb 19, 2024Jul 26, 2024Dec 31, 2024Jun 7, 2025Nov 12, 2025Apr 20, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
88.8-
2026.01
87.3-
2026.04
86.9-
2026.01
84.2-
2026.04
74.67-
2026.04
74.28-
2026.04
74.18-
2026.01
73.7-
2026.03
72.85-
2026.03
72.77-
2026.01
70.9-
2026.03
70.659.31
2026.01
69-
2026.01
68-
2023.09
66.6-
2026.03
66.5127.48
2026.01
65.1-
2026.03
64.9534.59
2026.01
61.2-
2026.04
60.3-
2026.04
58.9-
2026.04
58.7-
2026.04
58.6-
2026.04
58.2-
2026.04
56.5-
2026.04
55.59-
2026.04
54.2-
53.95-
2026.03
52.1390.66
2026.04
52.1-
2026.04
51.8-
2026.04
51.6-
2026.04
51.1-
2026.04
50.61-
2026.04
49.71-
2023.09
47.2-
2026.01
41.6-
2026.01
41.5-
2026.04
21.4-
2026.01
18.9-
2026.01
8.7-