the moment this clicked after reading it a few years ago its been my daily driver. that resource is a treasure

Agreed 100%. But for me it was many, many years ago. But this is my daily shoelace knot and it works amazingly well.

i have some fun experiments i'm doing with full virtualization middle ware of all sys calls for agents tools/shell commands/io, still far from daily driver, but allows me to do a very rich overlay / virtual file system tom foolery in place

I seem to have gotten an auto reject from your ATS before anyone technical could have read the application, which is odd because my career around building tools, libraries, compilers/interpreters plus associated experiments in applied math / performance engineering tensor algs fits the role to a T.

some of the sandboxing ive been playing with gives me the best of both yolo and like logic programming tier perms on llm actions in env. still not ready for prime time though ;)

“sit with it a moment” or similar phrases is one of the more common anthropic tells ive seen. ughh

Yeah, there are a lot of AI-isms in there. Also weirdly sometimes the entire "rythm" between sentences as well.

> The fear stacks. It compounds. And it creates political and regulatory blowback that lands on the entire industry, including the companies doing it well.

I am utterly repulsed by this now-pervasive staccato impact writing. LinkedIn is a quagmire for AI-generated copy.

Sit with it a moment - you know what I mean - it’s obvious - elementary knowledge… etc.

Are they clichés? Yes, they are. But AI certainly didn’t invent them.

It just amplified them, because even those who couldn’t put together an essay without AI now have the chance to write a coherent sentence.

AI is a ghost writer for people.

Call it extreme if you like, but I propose we hit it hard and hit it fast with a major; and I mean major; leaflet campaign.

I’ve seen people use that phrase long before AI.

Is this the new thing that’s going to happen now? People use phrases that have been around forever and people accuse them of using AI?

I’m glad I’m not going to college in this environment. How unfortunate and demoralizing it would be if I wrote an essay by myself in a college class and the professor thought I was using AI.

I haven’t been to college in a while. That was long before AI. I do have access to some of my old essays. Based on their tone and some of the wording I was using, I don’t doubt that some people would accuse them of being AI written were it not for the fact I wrote them many years ago.

Will college students now deliberately try to avoid certain common phrases out of paranoia of being accused of this?

> People use phrases that have been around forever and people accuse them of using AI

What's happening (in the most part) is that because people used that phrase long before AI, that its in the AI training set and being slopped back out at us.

its about overuse of rhetoric in a dilutive way.

sitting with something is for way more personal or emotionally intense shock than one ceo saying other ceos are lying sacks of shit about layoffs, albeit in ceo speak.

a republican party die hard from like the 70s or earlier would be so shocked and disgusted by the modern incarnation that would literally need to sit down for a while just to emotionally process their shock and disgust.

shitty ceos not owning their fuckups, not shocking

i cant find anything substantiated in the code that actually differentiates it from any other harness.

my fork of oh my pi that i have a lot of experiments in, is lterally designed to only work well with models that have decent reasoning levels, like deep seek models. check it out!

https://github.com/cartazio/oh-punkin-pi/blob/main/scripts/b... — thats the install script for after clone

fair warning: tis my dog food test bed as i build even fancier stuff

just use the same context cache and a separate completion assuming elastic resources

in my llm agent harness code i force all streamed reasoning to be turned into specially formatted visible to everyoneb msg blocks. it really does a nice job of fixing that while not caring about which provider for what

edit: pretty sure full cot is always available via api.

this sounds like a nice non prescriptive direction. jose is a pretty cool dude. pre covid we theoretically were gonna poke a defining a memory model of ghc haskell c— / hs / core , but life intervened

oh my, i see what youre saying. at this point youd hope everyone has realized that the best way to keep models more reliable is to force them to stay honest via very very string static typing as a feedback loop. bags of text with hyperlinks certainly fail that measure