2 ms·
I like alignment more it is pretty quantifiable and sometimes it goes against 'safety' because Claude and Openai are censoring models.
by antupis 3y ago
I like alignment more it is pretty quantifiable and sometimes it goes against 'safety' because Claude and Openai are censoring models.