Model-free RL is sample-inefficient because each sample tells the learner very little about what good on-policy trajectories look like until it gets close to convergence,and since it lacks a…

Posters: davidad. davidad

2 likes recorded by the source corpus.