It's crazy how people over-engineer the prompting part when all you need for latest models are a few examples after the prompt.

@xorswap_ Good for him tbh

@aidenybai iirc that's what nodejs does with --experimental-strip-types flag

@nearcyan @elysian_labs good luck with the launch

@atulit_gaur Bro said "how about you assemble with someone first?"

15 more minutes and already much better. vibe coding stays undefeated

RAG: Store docs in vector database, fetch relevant ones and add it to prompt "Fusion" RAG: Generate variations of query before sending "Agentic" RAG: Generate a dummy answer so that it's easier to find relevant docs. I think all this naming is webdevs discovering system design

@eddybuild "You're overreacting. That never happened."

There's a lot of Gemini-posting today. I'm aware 2.0 Flash is a good model, but haven't seen this many people shilling it before

@Aryvyo fairuz on tl

never kill yourself

@NizzyABI This is basically the logo of Bethesda

@buccocapital There are lots of charts for the last year that has a similar spike caused by Covid. If I had to guess I'd say this somewhat late reaction to the remote working trend that slowly died after quarantine

I know for a fact that Langchain is written in PyCharm. The codebase carries a distinct filth that could only emerge from the depths of a Professional Integrated Developer Environment

@shurensha @_MathAcademy_ I'm halfway through completing the ML math course. Before that, I had to finish the fundamentals 3 course, so my calculus and statistics are still fresh. The term hasn't started yet, so we'll see how useful it will be

@EsotericCofe it takes one broken config to lose even the God's most cracked soldier

@tekbog staff engineer for 20 years build