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

Compositional Reasoning on SugarCrepe++

75.2Average Performance

CLIPS (ViT-L/14)

42.75251.17659.668.024Sep 23, 2025Oct 23, 2025Nov 23, 2025Dec 24, 2025Jan 23, 2026Feb 23, 2026Mar 26, 2026
Updated 21d ago

Evaluation Results

MethodLinks
2026.03
75.276.666.659.444.3-----
2026.03
7579.775.355.244.2-----
2026.03
74.477.769.757.942-----
2026.03
73.475.464.955.642.1-----
2026.03
71.376.78652.141.3-----
2026.03
70.574.774.845.830.8-----
2026.03
69.974.37556.651-----
2025.09
63.7---------
2025.09
63.7----57.55292.17343.9
2025.09
61.8---------
2025.09
61.8----52.253.491.374.936.5
2025.09
57.5---------
2025.09
57.5----51.149.885.269.931.4
2025.09
56.3---------
2025.09
56.3----41.652.685.768.333.3
2025.09
54.7---------
2025.09
54.7----40.848.489.165.629.6
2025.09
54.1---------
2025.09
54.1----5345.181.258.232.9
2025.09
53.3---------
2025.09
53.3----39.147.485.362.432.5
2025.09
51.3---------
2025.09
51.3----41.839.580.756.638
2025.09
44---------
2025.09
44----29.548.670.15019.6