How Much Does It Cost to Train Atari Agents?

When I think about doing this in a loop, it seems like it could be maybe massively expensive. Like how do you think about the cost of these of these algorithms and when it actually makes sense economically or just when it makes sense to use these methods? Do you have a sense of that? Yeah, that's a great question. So for the few shot learning setting, it's not really hugely different from just training a like agent that can generally solve tasks of that flavor. There's the trick there is that these algorithms can generalize quite a bit. But I mean like yeah it's the most expensive Atari agents of that at a personalized level for sure.