shin

AI agents are deploying to prod, but can they autonomously find and patch unseen critical vulnerabilities? We introduce ZeroDayBench, a benchmark for evaluating LLM agents on proactive cyberdefense. Plus, a novel high-severity (CVSS 8.1) CVE we found partway through ... 👀

@nozomioai not sure what the actual retrieval process looks like under the hood, but either way it could be denser anyway great product, just rough thoughts!

@nozomioai yeah but /compact does a pretty bad job keeping what actually matters, save context goes the other way, full history, which is great but heavy on re-injection something in between would be sick -- summarize with the right rules, store dense, re-inject small

more than a month ago i thought about solving context sharing/handling techniques, and just found out @nozomioai already exists, which is an awesome tool in addition, it would be nice to have smth like compact: save -- summarizes before saving, keeps only signal (not obvious decisions, state, key paths, open questions)

@crypt0lake they start to get after the first visit in Europe

man it’s so funny how the japanese don’t know how good they have it

just be optimistic

@super_bavario > grandpa are you one of them? > ...yes, and i still don't know

>but how many features do these hud guys have grandpa? >i’m afraid there’s not a single person that knows that anymore little one, not even one

Aviro is introducing Ebla, a state of the art grounded reasoning model. In collaboration with HUD, the Aviro team built C⁴ — a benchmark for long-horizon tasks in corporate document sets. We evaluate four dimensions: Correctness, Completeness, Composition, and Citations. @aviro_ai post-trained GPT-OSS 120b to achieve SOTA performance, with a Pass@1 score of 25.4% and Pass@8 score of 37.1%.

@OkabeTech it's kinda random, mostly only in Abu-Dhabi, but i'm already home so all good!

@shfunc Bro 💀 I heard something about the state providing free hotels if you have a cancelled flight tho?

4 canceled flights, i'm going insane rn

@thegeneralist01 @niggachandesu type shit

@niggachandesu i’ve seen so many russian-speaking people, but close to none were russian also the propaganda machine is doing its job. social medias are banned in the mainland country.

@shayanshafii i really don't want @ludwigABAP to see it

Roy Lee is the closest thing Silicon Valley has to Kanye West

Four Waterloo students trained a SOTA real estate agent on hud in under 24 hours. 😪😪😪

@OkabeTech was some sort of vacation 😭

@shfunc Tf you doing in Dubai?!

@vladnineplusone слитый скрин новой евы

waiting on anthropic's new compact options because the current one is genuinely criminal in the meantime building my own context layer between the api and the agent

the bottleneck isn't context window size. it's that nobody's built a forgetting policy

cc creator is from my hometown, tf

@0xdiicell @ludwigABAP we're so back

yo, @ludwigABAP vertical tabs are live in helium

@OkabeTech yes please

@shfunc Delegate non important cognitive work to AI

work smart AND hard