Tootfinder

Opt-in global Mastodon full text search. Join the index!

No exact results. Similar results found.
@catsalad@infosec.exchange
2025-09-22 22:41:37

Those are my load bearing kitten photos.
🎨by extrafab

Four panel comic by Extra Fabulous Comics showing someone at their computer. First panel shows the person shocked while yelling, "Police hunting people for sport?!" Second panel shows them smiling and saying, "A chubby penguin! Haha" Third panel shows them all angry while screaming, "The President ate a baby?" Last panel shows them smiling again saying, "A kitten! Haha"
@stiefkind@mastodon.social
2025-09-21 09:02:59

»Die deutschen Versicherungsgesellschaften haben sich z. B. […] darauf verständigt, unterschiedliche Computer für Testläufe und für den echten Betrieb anzuschaffen. Es könnte ja sein, daß ein Programmierer, der seinen Testlauf mit echten Daten macht, auf die Idee kommt, diese Daten unbefugt zu kopieren. Eine Bandkopie, die […] alle fälligen Auszahlungen […] samt Name und Anschriften der Empfänger enthielte, wäre ein gefundenes Fressen für die Unterwelt.«
Aus: Werner Heine - Die Hacker …

@Mediagazer@mstdn.social
2025-09-23 07:45:45

Some left-leaning audiences are turning to independent outlets like Bulwark and Crooked Media, abandoning legacy media over their pro-Trump transgressions (Max Tani/Semafor)
semafor.com/article/09/21/2025

‘Look at the charts’: Democrats desert legacy media for new outlets
Angry viewers are trying to make legacy outlets pay for their pro-Trump transgressions by turning to competitors.

@arXiv_csLG_bot@mastoxiv.page
2025-09-18 10:20:21

Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision
Dulhan Jayalath, Shashwat Goel, Thomas Foster, Parag Jain, Suchin Gururangan, Cheng Zhang, Anirudh Goyal, Alan Schelten
arxiv.org/abs/2509.14234

Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision
Where do learning signals come from when there is no ground truth in post-training? We propose turning exploration into supervision through Compute as Teacher (CaT), which converts the model's own exploration at inference-time into reference-free supervision by synthesizing a single reference from a group of parallel rollouts and then optimizing toward it. Concretely, the current policy produces a group of rollouts; a frozen anchor (the initial policy) reconciles omissions and contradictions to…

@arXiv_mathAG_bot@mastoxiv.page
2025-09-23 08:23:50

Asymptotics of stability thresholds
Junyao Peng
arxiv.org/abs/2509.16458 arxiv.org/pdf/2509.16458…

Asymptotics of stability thresholds
We study asymptotic behavior of the stability thresholds of a big line bundle, and prove explicit bounds on the error terms. This answers Jin--Rubinstein--Tian's questions affirmatively. A key step in our proof is to show that the stability thresholds of a big line bundle can always be computed by quasi-monomial valuations. This generalizes Blum--Jonsson's result on the stability thresholds of an ample line bundle.

@curiouscat@fosstodon.org
2025-09-14 17:00:26

Photo of Angkor Wat (Cambodia) with a cow and a pig in the foreground by my father.
Angkor Wat is the largest Hindu temple complex in the world, situated at Angkor, Cambodia, built by King Suryavarman II in the early 12th century as his state temple and capital city.
#AngkorWat is a UNESCO World Heritage Site.
When I visited (decades later) tourists filled the fields instead of…

black and white photo of Angkor Wat (Cambodia) with a cow and a pig in the foreground.
@offenenetze@chaos.social
2025-11-16 20:19:39

Angreifer steigen über Zero-Day-Lücke bei Logitech ein | heise online
heise.de/-11080333

Angreifer steigen über Zero-Day-Lücke bei Logitech ein
Beim Hersteller für Computerperipherie Logitech gab es einen IT-Sicherheitsvorfall. Dabei konnten Angreifer auf interne Daten zugreifen.

@unchartedworlds@scicomm.xyz
2025-09-14 09:09:54
Content warning: LLM training frameworks, interesting

Interesting explanation of LLM training frameworks and the incentives for confident guessing.
"The authors examined ten major AI benchmarks, including those used by Google, OpenAI and also the top leaderboards that rank AI models. This revealed that nine benchmarks use binary grading systems that award zero points for AIs expressing uncertainty.
" ... When an AI system says “I don’t know”, it receives the same score as giving completely wrong information. The optimal strategy under such evaluation becomes clear: always guess. ...
"More sophisticated approaches like active learning, where AI systems ask clarifying questions to reduce uncertainty, can improve accuracy but further multiply computational requirements. ...
"Users want systems that provide confident answers to any question. Evaluation benchmarks reward systems that guess rather than express uncertainty. Computational costs favour fast, overconfident responses over slow, uncertain ones."
=
My comment: "Fast, overconfident responses" sounds a bit similar to "bullshit", does it not?
#ChatGPT #LLMs #SoCalledAI

@raiders@darktundra.xyz
2025-09-17 14:01:40

Raiders Notes: Geno Smith’s Confidence, Jordan Meredith Experiment, And More raiderramble.com/2025/09/17/ra

Raiders Notes: Geno Smith’s Mindset, Meredith’s Role, and Key Takeaways
The Las Vegas Raiders were left searching for answers after a deflating 20-9 loss Monday night, a game that ended without a single touchdown for the Silver and Black. Turnovers and missed opportunities, which stalled drives before they could gain momentum, caused the offense to sputter from start to finish. Quarterback Geno Smith, under heavy […]

@sean@scoat.es
2025-10-18 21:07:14

Remember when you could search for something and when there were no results, the computer on the other end would say “no results” instead of forcing you to scan through a bunch of “results” that aren't actually what you searched for in hopes that maybe your answer is there somewhere (but it's not)?
That was cool.