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

Long-context Visual Question Answering on Visual HayStack Long Window (test)

68.17Accuracy (50 images)

Base

40.048447.349254.6561.9508Dec 25, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.12
68.1762.5660.65-
2025.12
67.0463.3760.1724.5
2025.12
63.152.2843.083.6
2025.12
58.253.9150.2316.4
2025.12
56.2647.9642.36-4.7
2025.12
54.7351.0847.43-
2025.12
52.3448.0142.181.96
2025.12
52.245.0739.97-1.9
2025.12
50.3545.1240.12-11.7
2025.12
50.2247.841.67-
2025.12
41.1335.0531.25-23.26