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

Visual Reasoning on NLVR2 (test)

85.15Accuracy

SimVLM_HUGE

56.695664.082871.4778.8572Nov 3, 2021Jul 29, 2022Apr 24, 2023Jan 18, 2024Oct 12, 2024Jul 8, 2025Apr 3, 2026
Updated 13d ago

Evaluation Results

MethodLinks
2021.11
85.15
2021.11
83.98
2022.10
83.48
2024.03
83.2
2021.11
83.14
2024.03
83.08
2021.11
83.05
2024.03
82.85
2021.11
82.47
2024.03
82.42
2022.10
81.81
2021.11
81.77
81.72
2021.11
81.47
2024.03
81.23
2025.12
81.2
2024.03
81.13
2021.11
80.5
2025.12
80.5
2024.03
80.01
2021.11
79.98
2026.04
79.67
2022.10
79.26
2024.03
79.22
2022.10
78.36
2025.12
78.2
2021.11
78.05
2026.04
77.64
2024.03
77.61
2024.03
77.61
2022.10
76.79
2022.11
76.3
2021.11
76.13
2022.10
76.1
2025.12
75.2
74.3
2022.10
73.93
2025.12
73.6
2024.03
73.55
2021.11
72.2
2024.03
68.76
2025.12
68.3
2022.11
62.4
2022.11
61.8
2022.11
61.3
2024.03
57.79