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

Response Diversity on Anthropic HH-RLHF

82.5Preference Coverage

EvoPref

66.48470.64274.878.958May 10, 2026
Updated 21d ago

Evaluation Results

MethodLinks
2026.05
82.50.297110.84
2026.05
78.20.32714.10.82
2026.05
76.90.34215.50.8
2026.05
72.50.36118.10.71
2026.05
700.38920.6-
2026.05
67.10.41423.3-