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

Preference Prediction on VideoGen-RewardBench

76.81Score @ Threshold 99

PRISM

46.878854.649462.4270.1906Jun 18, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
76.8176.4475.7874.7972.6369.8760.3674.12
2026.06
75.8363.7252.9949.3750.1950.349.8677.38
2026.06
74.5174.4273.4372.6870.0167.5659.5672.93
2026.06
67.8159.8851.8752.5551.8451.7450.6472.38
2026.06
63.9863.763.1362.360.4658.1650.2561.68
2026.06
63.1963.0862.8962.6460.3158.2451.561.8
2026.06
63.153.0244.1741.642.1942.341.9364.39
2026.06
62.0761.9961.1360.558.2856.2649.660.76
2026.06
56.7652.8926.4916.8617.8717.8217.258.27
2026.06
56.4349.8343.1643.7343.1443.0542.1360.23
2026.06
52.6152.5352.3652.1550.2548.5443.2551.44
2026.06
48.0344.3627.1920.6821.1721.3121.0249.23