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

Visual Reasoning on NLVR2 (dev)

82.5Accuracy

MADTP

56.156862.995969.83576.6741Mar 5, 2024Jul 12, 2024Nov 18, 2024Mar 28, 2025Aug 4, 2025Dec 11, 2025Apr 20, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2024.03
82.5
2024.03
82.48
2024.03
81.97
2024.03
81.92
2024.03
80.67
2024.03
80.33
2024.03
79.5
2026.04
79.13
2024.03
78.28
2024.03
78.08
2026.04
77.16
2024.04
77
2024.03
76.89
2024.04
75.5
2024.04
73.8
2026.04
73.7
2024.03
72.85
2026.04
71.7
2026.04
71.1
2024.03
68.71
2024.03
57.17