Is There a Unified Algorithm in Meta Rl?

So in regular deep rl we've seen an explosion of algorithms and uh but recently we've seen the dreamer and recently dreamer version three from dandjar halfner at all that beats a lot of algorithms without tuning. That suggests there's some maybe some pruning or convergence of the state of the art family tree of our algorithms is maybe an order. i guess dreamer v3 already will solve a huge chunk of meta rl problems so it's already a good start um but i think that really there are different problem settings with pretty unique demands in the meta learning setting right? So if you have a very narrow distribution of tasks and you don't have to do much extrapolation