@OortCloudAtlas@AlexKrusz On the contrary, the larger the model the more likely it is to converge to buddhist discussions (in one antrophic study).
I would also expect larger models to be better grasp those (non-)concepts.
I’m curious, do you see improvements going ie from sonnet->opus->fable?
@OortCloudAtlas Im a big fan of yours, but honestly I think you need to let that topic go. I don’t think you talking about it that way improves the world, neither short term nor long term.
@OortCloudAtlas I believe having the raw q&a available would already be a valuable resource. I imagine using it with AI as a search tool. Instead of the AI having to come up with a answer itself, it can refer to similar questions of other students. Thus the answer will be trustworthy.
@omouamoua What is the difference then to just combining the two batches?
(So the gradient would be the average and remove ‚non-generalizing‘ signal anyway)
@OortCloudAtlas In times of change I believe it’s best not to cling too much to old concepts. Instead, personally I interpret our time more akin to birthing, where the process is challenging yet the result is utterly beautiful.
@DavidSHolz @mr_tjneo @midjourney I love midjourney but I think it has lost its edge. Maybe the best course of action going forward would be to partner with a different model provider and use those models in conjunction to the midjourney model. Midjourney for aesthetics and the artistic process, else the other