Reaction
Critic-only reinforcement learning in continuous control
A post calls for simpler RL and welcomes what its author sees as momentum for critic-only approaches.
TLDR
The author calls for simpler reinforcement learning and says it’s encouraging to see momentum for critic-only RL in continuous control. They link a paper they describe as a better-focused expansion of their previous work on growing Q networks.
Combined views
93
1 Source, first seen ago