Remote

udo m. rader ☕ 🇪🇺 🇺🇦 🐧

@riaschissl@sigmoid.social

mastodon 4.7.3

Open source pragmatist, >25ys in IT, family guy, father of three, loves the mountains, CTO, alumnus Uni Salford & Uni Krems. #PhD student.

Remember: there is no such thing as too much ☕

Pro-European 🇪🇺 #LGBTQ+ ally 🏳️‍🌈

Money is poverty.

What you'll mostly hear from me is about
#Infrastructure #DevOps #AI #GitOps #FreeBSD #Linux #k8s #java

... and I'm also very fond of
#running #TrailRunning #UltraRunner #skiing #startrek #discordia #pastafari

Plus ... loads of random ideas 😎

:x!

0 Followers

0 Following

22 Posts

Joined June 03, 2024

Home:

Austria, EU

github:

https://github.com/daudo

pronous:

all pronouns welcome

born at:

327 ppm

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 6mo ago

I’ve been using #Forgejo as a #Git repo server for a few days now for some of my personal & PhD projects and experiments.

And what can I say: I’m thrilled! I’m very familiar with #gitlab, but Forgejo has nothing to fear from the comparison! Of course, there's no feature parity, but it seems Forgejo/Gitea have simply focused on the essential aspects, and that works perfectly. Excellent performance, low resource footage, good CI/CD. On top of that, it’s fully FOSS, no ifs or buts!

#selfhosting

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 6mo ago

It's that time of year when #running in the mountains you can experience it all within a matter of just a few metres: spring, slush and the depths of winter.

After Easter, the last ski resorts in the area close, and then it's only about ski touring or #TrailRunning. Or both 🤷‍♂️

Happy Easter, Merry Ostara, or simply a lovely few days off to you all!!

#garmin #beatyesterday @running@a.gup.pe

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 4mo ago

Replying to

@nixCraft@mastodon.social and now add this recent mind blowing statement by #GitLab's CEO: "Software will be built by machines, directed by people. AI is the substrate on which future software gets built. Agents will plan, code, review, deploy, and repair. Humans still own the judgment that matters most: architecture, deep understanding of the customer problem, the tradeoffs that require taste." As much as I like #AI as a technology, I am disgusted by the openly displayed greed https://about.gitlab.com/blog/gitlab-act-2/

GitLab

GitLab Act 2

A letter to our customers and our investors.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 2mo ago

Replying to

@briankrebs@infosec.exchange at the end of the talk, the dilemma is described nicely: the capabilities to launch sophisticated attacks are plenty (and they are remarkably smart, to put it mildly), but the defences fall far short on almost every level. Automated hacking, but no serious (automated) defences - apart from pulling the plug …

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 5mo ago

Replying to

@glynmoody@mastodon.social IDK ... commercial VPNs are big in marketing "privacy" when indeed they often do quite the opposite. Many if not most of those VPNs are operated by quite shady (and state affiliated) businesses, and trusting them with anything "private" seems to be quite strange.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 2mo ago

Replying to

@lehtimaeki@snapp.social that really looks like an interesting project! However, their roadmap is extensive, and it seems - at least according to the roadmap - that they still have a long way to go. Many basic features are still missing (VO2max, hill running with an altimeter, etc.). https://roadmap.unawatch.com/roadmap

roadmap.unawatch.com

Upvoty

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 4mo ago

Replying to

@jik@federate.social I couldn't agree more. "Crap in, crap out" comes to mind. #GenAI is good at repeating the patterns it has been trained on, but it has no idea whether a pattern is good enough, let alone how to generate a creative and useful one.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 6mo ago

Replying to

@catsalad I appreciate the fact that this is a multiple choice question. Because that's exactly how feline science is supposed to work. One of the many things we can learn from Schrödinger's cat!!

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 6mo ago

... and then you investigate an issue in your tier 0 infrastructure, only to discover that #ansible has been logging heaps of credentials for weeks because 1. some crazy guy didn't use "no_log: true" properly and, just as grave, 2. nobody noticed until now ...

#WTF #DevSecOps #GitOps

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 6mo ago

AI discourse on Mastodon feels increasingly binary: either total utopia or absolute doom.

I'm curious if this polarization reflects reality or just the loudest voices. I'm running a quick pulse check on how YOU all actually feel about LLMs and AI talk in your timeline.

No right answer, just honest data.

Results will be followed by a breakdown of where I stand on this spectrum.

Boosts welcome

#AI #Mastodon #TechDebate #GenAI #ArtificialIntelligence

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 4mo ago

Replying to

@iftas@mastodon.iftas.org thank you nevertheless!! A real pity that the admins don't seem to be around.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 6mo ago

Replying to

@Johan_Barelds congrats!! This looks like a particularly tough trail ... for me running in sand always feels like glue pulling down my shoes, almost double the effort! Enjoy the well deserved rest!!

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 4mo ago

Replying to

@lehtimaeki@snapp.social What a wonderful documentary - it captures my own personal view of Ultras so perfectly. Heroes aren't just those who succeed (perhaps even fast), but also all those who give it their all, even if they end up failing. But either way, they definitely learn a great deal about themselves. Humility.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 5mo ago

Stakeholders claim #NuclearPower is the #EU's only path to reliable, fossil-free supply.

Right, let's look at yesterday's prices. There's simply not enough electricity without nuclear stations!! Solar, wind, and others just don't provide enough. Who can afford -40 cents/kWh (yes, minus 40!!) 🤡

#RenewableEnergy #GreenDeal #ClimateChange

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 8mo ago

Replying to on social.tchncs.de

@oli@social.tchncs.de I have to disagree on one point: it takes surprisingly little hardware resources to run #GenAI efficiently for say 20 ppl. The hardware required for this is now quite affordable for a reasonably sized company as well as for enthusiasts (excluding the current crazy RAM prices). And decentralised operation neither requires a nuclear power plant nor an ocean for cooling, it works perfectly well with renewable energy. Decentralisation is the key, one could perhaps even call it democratisation.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 5mo ago

Replying to

@hetzner@mastodon.hetzner.social as much as I value the certification, when I look at this picture, I immediately ask myself: Why aren't the roofs covered with solar panels?

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 7mo ago

Replying to

@dazo First, please don't get me wrong: Like anyone else, I don't want my data to be sold, and at the very last by the browser I've been using as my daily driver for anything internet for years. But the question I can't find a viable answer for is: How can Firefox become a sustainable organization? Selling our data is a no-go, that's for sure. But how do they make the money required to not only maintain but also invest in Firefox?

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 8mo ago

Replying to

@dzervas@infosec.exchange @oli@social.tchncs.de the main question is: what is good enough for a given task? As we know since GPT 4.5, pure parameter count isn't the main enabler for "better" LLM results. And then we have Unsloth's Dynamic 2.0 quantization, which is extremely helpful to bring down the VRAM requirements while still keeping the accuracy at a very high level. So yes, pack two or three RTX 6000 Blackwell GPUs into a decent server, and you're good to go at a decent price.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 8mo ago

Replying to

@dzervas@infosec.exchange @oli@social.tchncs.de take the cheapest Claude Max subscription for 100 per month, this sums up to 24k per year for 20 devs. And if you just take the ordinary team subscription, it's more than 7k per year. So the economical sense is there, plus not having to worry about what happens to your data. And in a corporate environment, replacing hardware every 5 years is not unusual, if not the norm.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 2mo ago

Replying to

@Drachenstreichler@mas.to zwecks "Schwarmidiotie" gebe ich Ihnen mehr oder weniger recht. Weniger va. deshalb, weil genau das das Hauptargument von Typen wie Thiel und seine kruden Thesen ist, dass Demokratie nicht funktioniere. Ich hege die naive Hoffnung, dass diese Idiotie großteils heilbar ist. Die Sache mit AI und den DCs sehe ich weniger negativ, weil in diesem Fall der Kapitalismus zuschlagen wird: AI, so wie sie derzeit gepusht wird, rechnet sich schlichtweg nicht. Diese Blase wird platzen.

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 4mo ago

Replying to

@iftas@mastodon.iftas.org yes, I've also seen this on the instance I'm on, and it's getting worse by the day. The local feed has become more or less unusable. Unfortunately the admins appear to be unresponsive. Any ideas what to do about it? https://sigmoid.social/@riaschissl/116629296136277143

sigmoid.social

udo m. rader ☕ 🇪🇺 🇺🇦 🐧: "@thegradient don't want to bug you or anything , …" - Sigmoid Social

udo m. rader ☕ 🇪🇺 🇺🇦 🐧 @riaschissl@sigmoid.social

· 8mo ago

Replying to

@oli@social.tchncs.de Yes, using our existing DC infrastructure may not be as efficient as once again surrendering our sovereignty to the same tech bros we regularly complain about. And I would argue that more than 90% of today's typical workload is inference rather than training. And to your last point: give any two software devs the same task and you'll get two different results. Have them do it again and you'll get another set of different solutions. Absolutely non-deterministic. No different with LLMs.