Environmentalist protestors do something sensible challenge, impossible.

Does anyone know of a large repository of prompt engineering somewhere??

Has anyone made a local model purely focused on prompt engineering?

i.e., I say what I want + whatever data input I need, and it creates the best prompt for me?

What LLMs can never do!

Very good read ⬇️

buff.ly/4ddvlcs

What can LLMs never do?

Like, why can’t LLMs do Wordle? Interesting, long, nerdy post on the current and future limits of large language models strangeloopcanon.com/p/what-can-llm…

As I found out throughout my research, one of the Japanese 'shinise' secrets is an extreme dedication to doing only one thing.

rohit concluded the same. in his essay, The Price of Immortality.

strangeloopcanon.com/p/the-price-of…

This is a really good package from UK's safety institute on running evals. And it's open source too! Excellent work.

Demis sounds confident

Kids definitely are getting smarter. You don't develop a meaningful life philosophy by going to classes in college. It's the result of a meaningful life well lived, which is almost antithetical to college these days.

Good thread!

RicG It's just *really easy* to squint and abstract yourself into a place where 'the Henry Adams curve implies we'll have accelerating returns towards doomsday drills' sounds reasonable if you throw out all engineering details and plot everything log log with a fat magic marker.

Who needs bread and circuses when you have brain worms

All compression is destruction