List your company

Button Press on Meta-World Button Press human preferences

57.2Success RatePAWS
44.247.57550.9554.325Jun 10, 2026

Evaluation results

MethodLinks
PAWSArchitecture=Transformer2026.0657.2
PAWSArchitecture=MLP2026.0656.2
Pref Trans.2026.0655.2
P-IQL2026.0653.2
CPL+KL2026.0649
IPL2026.0647.8
CPL2026.0647.5
BC2026.0644.7