Oooh this looks fun!

My first course when k moved to France was “analysis réelle et complexe”. I think first lecture we went all the way up to Baire category theorem. All that to say the French way of doing math is … different. I still enjoyed it, but still don’t grok measure theory :)

Seconded and thirded! You absolutely have to attend this.

Wouldn’t a RAG based approach work better: compare the user query to functions in the api spec and then let the LLm choose between the results returned ? Or am I missing something?

Couldn’t agree more. The sloppiness and just the lack of pace at the beginning of the second half. It would have been better if we were 1-0 down because we wouldn’t have left the foot off the gas till the game was buried. Geez!!

Never understood this as a line of critique: sure it uses GPT-4 but building a product around a model is no joke. Thinking through the use cases, doing the engineering and then shipping the end product is not simply typing something in chat gpt.

The game was there to be won. And I feel we I.e. Liverpool didn’t take our chances.

Twitter doing its job. Follow good people here and you get to book recommendations for free. :)

The prize looks like an awesome book. Lipsa Nag might be interested too.

This idea of using LLMs as key components within traditional ABMs was the premise of my ongoing work on embodied agents. Read more here: dhruv-sharma.ovh/post/talk-ccs-…

Basic structure (in code) : github.com/dhruvshrma/emb…

Good to know that I am not the only one making my variable names very explicit (and match the model notation that I have setup)

Can AI be our friend? New post out now: dhruv-sharma.ovh/post/ami-not-a…

❤️