
@mattpocockuk There are lots of discussion against structured output as it impedes LLM quality. The most robust implementation I have seen is unstructured output + another LLM to structure the first response
English
Andi Partovi
44 posts

@andiPartovi
Always Building










Any Stanford founders on here? I’m hosting a fundraising workshop on campus Wednesday. Reply if you want an invite or for the next one in SF



going to invest $1,000,000 into the batch. Starting Feb 24th for 4 weeks, dropping first round of acceptances on the 14th. All happening in SF at our founder lab.


