List your company
ResearchFormerly WizWand

Reinforcement Learning on MountainCar (Average Episode Reward)

199.99Avg Episode RewardDQN
-122.3996-38.702344.995128.6923Nov 2, 2023Apr 10, 2024Sep 17, 2024Feb 24, 2025Aug 3, 2025Jan 10, 2026Jun 20, 2026
Updated 4mo ago

Evaluation results

MethodLinks
DQNNumber of runs=10, Aggregation protocol=averaged across all training iterations, Source=Best SB32026.05199.99
ProPSNumber of runs=10, Aggregation protocol=averaged across all training iterations2026.05199.31
ProPS+Number of runs=10, Aggregation protocol=averaged across all training iterations2026.05195.81
R2PONumber of runs=10, Aggregation protocol=averaged across all training iterations2026.05147.84
DSP2023.110.9911
Human BestMethod Type=Human Baseline2026.060.99
ARTSBase Model=o32026.060.9573
DDPG2023.110.9536
Regression2023.110.9516
SAC2023.110.9468
ESPL2023.110.9402
A2C2023.110.9397
TD32023.110.9387
ACKTR2023.110.9379
PPO2023.110.9376
TRPO2023.110.936
ARTS*Base Model=Qwen 4B, Test-time trained=true2026.060.9194
MLEvolveMethod Type=Prior Works2026.060.8484
AIRAMethod Type=Prior Works2026.060.8082
ARTSBase Model=Qwen 4B2026.060.5665
LinearMethod Type=Prior Works2026.060.4544
Average Human2024.05-70
DDQN2024.05-100
FDQN2024.05-107
DQN2024.05-110