Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Role-playing Multiple-choice Evaluation on RoleAgentBench

51.02Friends Accuracy

Sparse

17.06425.879534.69543.5105Feb 6, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.02
51.0253.9760.5361.7670.83
2026.02
45.9154.7663.1655.8862.5
2026.02
41.845055.265054.17
2026.02
39.848.4152.6352.9450
2026.02
30.6147.625034.0945.83
2026.02
23.4745.2444.7434.0945.83
2026.02
18.3742.0642.1129.5541.67
2026.02
18.3742.0642.1129.5541.67