Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Off-Policy Evaluation on RiverSwim mostly-left policy, T=50

55.7Qπ(1, 0) Coverage

Model-based bootstrap with pivot CI

-2.22812.81127.8542.889May 12, 2026
Updated 21d ago

Evaluation Results

MethodLinks
2026.05
55.757.34455.755.9565035.944.5
2026.05
53.345.923.453.35344.929.414.522.7
2026.05
53.148.353.453.151.951.451.951.152
2026.05
52.3-49.852.351.451.152.151.850.1
2026.05
52.348.545.352.35250.451.446.143.2
2026.05
51.144.431.251.150.647.540.432.532.4
2026.05
50.651.251.950.649.950.551.351.752.1
2026.05
50.546.615.450.5514842.632.314.2
2026.05
5047.247.55049.548.24947.446.2
2026.05
49.750.249.449.749.650.350.249.548.8
2026.05
49.546.651.449.549.248.749.349.847.3
2026.05
49.34950.149.350.149.450.749.951.2
2026.05
48.446.649.848.447.747.747.849.849.7
2026.05
47.946.24847.947.546.947.54848
2026.05
40.728.47.340.739.133.12314.37.7
2026.05
38.929.622.938.936.429.78.73.623.8
2026.05
31.837.524.931.831.735.135.431.325.5
2026.05
29.330.320.629.33129.632.230.820.3
2026.05
000000000
2026.05
000000000