ResearchBenchmarksContinuous Control on OpenAI Gym Reacher (held-out λc ∈ {2, 3, 4})Follow-6.093RewardK_θ - Û-6.73572-6.56886-6.402-6.23514Mar 2, 2026Evaluation ResultsMethodMethodLinksRewardK_θ - Û2026.03-6.093K_θ2026.03-6.158NLOT2026.03-6.173CFM2026.03-6.251K_I2026.03-6.307K_I - Û2026.03-6.397MFM2026.03-6.561Direct2026.03-6.711