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

Human Preference Evaluation on EgoDex and DreamDojo-HV novel out-of-distribution (eval)

73.5Physics Correctness

DreamDojo-14B

62.0665.036870.97Feb 6, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
73.572.55
2026.02
72.565.53
2026.02
62.563.45