AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
How to Converge Algorithmic Reasoning With G and N's
i would really like these approaches to be reconciled going forward in the sense that uh like i don't see them as going one without the other if that makes sense so on one side and i'm gonna invoke the same principles i mentioned during our during our mlst episode you know daniel conman's book system one and system two right i think you cannot have one without the others. These kinds of specifically tailored prompts might enable the model to take things a step or two further but it's always like it's kind of in in spirit it's the same thing as algorithmic reasoning because we teach a model to execute an algorithm by forcing it to imitate the algorithm step-by-step here