And the fact they used a Chinese model, because none of the frontier models from very highly valuated top US companies support their very common and essential use case.
I also thought this was hilarious. The very "safety" measures these models have prevented them from doing... Something that is designed to increase safety?
And no similar sized big model is even public, which is super important to note.
There’s an article from yesterday I think it was stratchery where they say it’s also because the Chinese open source models are better because they don’t have to play by the no-distilling rules that the western models have to honour.