Fully agreed, and my hope is as open models grow and change, that getting some amount of this working on Pro-sumer hardware will be more attainable. But certai

Fully agreed, and my hope is as open models grow and change, that getting some amount of this working on Pro-sumer hardware will be more attainable. But certainly seems like we are a few years away from that, sadly. Am I also screwed in being able to train my own small model or adjust another one with such a non-workhorse PC?