List your company

Offline Reinforcement Learning on D4RL Gym v2 (test)

105.9Score (HalfCheetah, Medium-Expert)MBOP
38.19655.77373.3590.927Jun 17, 2022Jan 5, 2023Jul 26, 2023Feb 14, 2024Sep 3, 2024Mar 24, 2025Oct 13, 2025

Evaluation results

MethodLinks
MBOP2022.06105.955.170.244.648.84142.312.49.747.8—
D-QL2025.1096.8111.1110.151.190.58747.8101.395.587.9—
IDQL-A2025.1095.9108.6112.75165.482.545.992.185.182.1—
MISA2022.1094.7109.8109.447.467.184.145.698.686.2—742.9
χ-QL2025.1094.2111.2112.748.374.284.245.2100.782.283.7—
BooT2022.0694102.3110.450.670.282.946.592.987.681.9—
GTP2025.1093.8112.2114.253.990.389.550.8101.794.289—
QGPO2025.1093.5108110.754.1988647.696.984.486.6—
OneStep RL2022.1093.4103.311348.459.681.838.197.549.5—684.6
BDM2025.1093.2104.9111.15798.487.451.692.789.287.3—
10%BC2022.1092.9110.910942.556.97540.675.962.5—666.2
DTvariant=GPT22022.0691.8110.9914467.481.340.394.472.780.1—
CQL2022.0691.6105.4108.84458.572.545.59577.277.6—
CQL2022.1091.6105.4108.84458.572.545.59577.2—698.5
CQL2025.1091.6105.4108.84458.572.545.59577.277.6—
ARQ2025.109111010945618142816676.2—
TD3+BC2022.1090.798110.148.359.383.744.660.981.8—677.4
TTvariant=Re-train2022.0690.567107.644.164.981.636.590.850.369.4—
TTvariant=+S4RL (All)2022.0687.592.6100.546.753.980.243.176.152.171.1—
DTvariant=Original2022.0686.8107.6108.142.667.67436.682.766.674.7—
DT2022.1086.8107.6108.142.667.67436.682.766.6—672.6
IQL2022.1086.791.5109.647.466.378.344.294.773.9—692.6
IQL2025.1086.791.5109.647.466.378.344.294.773.977—
TTvariant=+S4RL (Last)2022.0684.396.9—46568040.272.861.170.9—
C-AC2025.1084.3100.4110.469.180.783.158.799.779.585.1—
TTvariant=Reproduce2022.0681.466.299.146.167.171.143.186.466.970.1—
BC2022.0659.979.636.643.163.977.34.327.636.947.7—
BC2022.1055.252.5107.542.652.975.336.618.126—466.7
AWAC2022.1042.855.874.543.55772.440.537.227—450.7
TTvariant=Original2022.0640.8106102.242.879.178.344.199.479.472.6—