List your company

Open-ended Dialogue on Anthropic-Harmless ID

68.1Win RateCausalRM
33.2642.30551.3560.395Jan 29, 2026

Evaluation results

MethodLinks
CausalRMOpponent=SFT2026.0168.123.88.1
CausalRMOpponent=Standard RM2026.0155.930.913.2
CausalRMOpponent=InfoRM2026.0138.537.224.3
CausalRMOpponent=GoalRM2026.0134.644.121.3