[rboyd]

Great work! There should be a way for entities to crowdfund model training. Can a model like this be partially evaluated during training time and save through early stopping?

What are the best papers/resources on sota long-horizon RL?

Thanks.