List your company

User Simulation Quality Assessment on Turn-level Human Evaluation Set adversarial (test)

168Win RateUserLM-R1
62.9690.23117.5144.77Jan 14, 2026

Evaluation results

MethodLinks
UserLM-R1Opponent=DeepSeek-R12026.011682042
UserLM-R1Opponent=Gemini-2.5-Flash2026.011424038
UserLM-R1Opponent=UserLM-R1 (w/o RL)2026.016711142