Continual Reinforcement Learning
Benchmarks
DatasetSOTA methodMetricTrendResultsLast updated
96Average Performance
5
Mar 4, 2026
90Average Performance
5
Mar 4, 2026
0.722C1 Forward Score
3
Mar 13, 2026
0.039Forgetting
3
Mar 13, 2026
0.194Forgetting
3
Mar 13, 2026
0.521C1 Final Score
3
Mar 13, 2026
-0.029Forgetting
3
Mar 13, 2026
-0.012Forgetting
3
Mar 13, 2026
HumanoidBench First Five Tasks Stand Walk Pole Slide Run (final checkpoint after five-task sequence)
25Average Success Rate
2
May 7, 2026
0.41Mean
2
May 7, 2026
29AP Gain
1
Apr 16, 2026
6.9AP Gain
1
Apr 16, 2026