Has anyone done a well executed lesswrong finetune?
Couldn't find anything decent and i think this might be a good experiment.