Continual Reinforcement Learning on MinAtar benchmark leaderboard · SOTA2 Research

Continual Reinforcement Learning on MinAtar

14.54Breakout ScoreFAME

Evaluation Results

Method	Links
FAME2026.03		14.54	18.72	1.69	0.16	0.72
LargeBuffer2026.03		10.71	3.24	1.16	0.16	1.65
Finetune2026.03		10.62	4.95	0.89	0.13	1.26
MultiHead2026.03		6.85	3.26	0.94	-0.01	1.25
Reset2026.03		6.51	3.29	0.74	0	1.31
PT-DQN2026.03		0.39	0	0	0.07	1.64