ResearchBenchmarksReward Modeling on LLMBar (test)Follow0.2039Test MSE (Table)FCPS0.203040.2088450.214650.220455May 31, 2026Evaluation ResultsMethodMethodLinksTest MSE (Table)Test MSE (Scalar)Test Delta (Δ)Test Paired 95% CI (Lower Bound)FCPSnM=300nM=3002026.050.20390.18260.02140.0186FCPSnM=200nM=2002026.050.21040.19140.01890.0123FCPSnM=100nM=1002026.050.21470.19720.01760.0126FCPSnM=50nM=502026.050.22130.21110.01020.0033FCPSnM=20nM=202026.050.22540.22120.0042-0.0036