

I don’t understand why anyone would think this would ever be possible for LLMs.
Yes, as I understand it, inference requires strictly less computational capacity than training; it’s just that a training setup can be turned around and run inference sessions massively in parallel. Just because you can run inference doesn’t mean you can accomplish a training run. Unless there’s some rationalist super-computer-science I’ve been missing this whole time…

I have more hope for putting serious pressure on my middle-of-the-road Dem rep. The district here only really turned Dem around 20 years ago thanks to the influx of Bay Area bedroom-community commuters, but there are still plenty of families around who immigrated as ag workers. So he already felt a degree of pressure, but now I feel like urging him to get on board with properly punishing the very worst Trump excesses might (might!) not be an empty gesture anymore.