Using Meta Rl With Offline Data

The paper also discusses using meta rl with offline data can you say a couple things about that. The idea with the offline inner loop is that we're already trying to do few shot learning so at the limit of this it's like you're giving some data upfront and you actually never have to do any sort of exploration in your environment. Someone from our lab zang the recent paper trying to generalize across different robot morphology with different action spaces um and there he's using hyper network which is also other work we've done in our lab hyper networks in meta rl. So kind of what is to help constant and what changes between these uh usually the action space is held constant the safe space