@prologic@twtxt.net Very rarely. And if I/we do, then it’s by train or by car. 😅

@lyse@lyse.isobeef.org @bender@twtxt.net Pfft, they want folks to relocate to Sydney. Fuck that 🤣 Sydney is a bit like San Francisco, I’m not actually sure which is worse. Fuck’n expensive as hell, the only palce you’d be able to afford to buy or rent is at least ~2hrs out of the city by public transport (i.e: train) and by that time you’ve just pissed your life down the toilet, because you’d be expected ot work a 9-10hr day + 2-3hrs of travel each way, buy the time you factor in having to wake up super early to get ready to travel in to work, you basically have zero time for anything else, let alone your ufamily,

Fuck that.

@movq@www.uninformativ.de Oh shit! :-( Time to switch companies. If you found something, please let me know. This hype train is derailing here as well.

A mate just sent me Microsoft’s magnificent master piece diagram regarding the end of life of Windows 10: https://support.microsoft.com/de-de/windows/windows-10-support-wurde-am-14-oktober-2025-eingestellt-2ca8b313-1946-43d3-b55c-2b95b107f281

That’s what you get for training with zalgo. :-D Of course, this isn’t even proper German.

In case they fix it, here’s a screenshot of the enlarged frontal crash: https://lyse.isobeef.org/tmp/win10eol.png

We had some gray soup with the occasional fine rain with strong wind gusts. Despite the bad forecast we took the train to Geislingen/Steige and strolled up to the Helfenstein castle ruin. All the colorful leaves were so beautiful, it didn’t matter that the sun was behind thick layers of clouds.

We then continued to the Ödenturm (lit. boring tower). By then the wind had picked up by quite a bit, just as the weatherman predicted. We were very positively surprised that the Swabian Jura Association had opened up the tower. Between May and October, the tower is typically only manned on Sundays and holidays between 10 and 17 o’clock. But yesterday was Saturday and no holiday. The lovely lady up there told us that they’re currently experimenting with opening up on Saturday, too, because there are some highly motivated members responsible for the tower.

We were the very first visitors on that day. Last Sunday, when the weather lived up to the weekday’s name, they counted 128 people up in the tower. Very impressive.

The wind gusts were howling around the tower. Luckily, there are glass windows. So, it was quite pleasant up in the tower room. Chatting with the tower guard for a while, we got even luckier: the sun came out! That was really awesome. The photos don’t do justice. As always, it looked way more stunning in person.

Thanks to all the volunteers who make it possible to enjoy the view from the thirty odd meters up there. That certainly made our day!

After signing the guestbook we climbed down the staircase and returned to the station and headed back. The train even arrived on time. What a great little trip!

https://lyse.isobeef.org/wanderung-auf-die-burgruine-helfenstein-und-den-oedenturm-2025-10-25/

@bender@twtxt.net I’ve been cutting back too 😅 Trying to avoid drinking (with alcohol) 24hr before fixtures (🏓) and before training 🤣

A cargo train ripped off several hundred meters of catenary and during construction they found a WW2 bomb. If I had gone to the office today, I would not have made it home for two reasons. https://www.swr.de/swraktuell/baden-wuerttemberg/stuttgart/bombenfund-in-stuttgart-untertuerkheim-100.html

@eldersnake@we.loveprivacy.club Yeah for sure! The thing that annoys me about a lot of this, is the sheer fact you can’t really self-host let alone self-train these things I’ve been playing around with AI at home over the past few months and building my own neural networks from scratch (in Go) with genetic algorithms on a few tasks and training sets, but man it’s hard™ 🤣 I feel like we’re doing something wrong here…

@movq@www.uninformativ.de I also don’t think that I’m a particularly good speaker. :-) The workshop model is a good idea, I like that.

Yeah, it’s really good fun. I can highly recommend it. This is also a good way to train (new) developers to think like attackers, how to break in, destroy something or raise awareness of some classes of bugs. Then you can avoid them next time. It’s surprising to me what vulnerabilities come up during this event every time. So, absolutely worth it, win, win.

@lyse@lyse.isobeef.org

They’re all talks, not real hands-on trainings like you did.

I love listening to good, well-structured talks. Problem is, not everybody is a good speaker and many screw it up. 🥴 I’m certainly not a great speaker, which is why I gravitate more towards “workshops”, in the hopes that people ask questions and discussions arise. Doesn’t always work out. 🤣 At the very least, I almost always have some other person connect to the projector/beamer/screenshare and then they do the stuff – this avoids me being wwwwaaaaaaaaayyyy too fast.

We are usually drowned in stress and tight deadlines, hence events like today are super rare … We used to do it more often until ~10 years ago.

Once a year the security guys organize a really great hacking event, though.

Oh dear, I’d love to participate in that. 🤯 That sounds like a lot of fun. (Why don’t we do this?!)

@movq@www.uninformativ.de Interesting internal education sessions are way too infrequent here as well. There are a bunch of “knowledge transfer” meetings actually, but 90% of the topics already sound totally boring to me. The other 9% talks turned out to be underwhelming, sadly. I only attended a single one where it was delivered what has been promised. They’re all talks, not real hands-on trainings like you did.

Once a year the security guys organize a really great hacking event, though. Teams can volunteer to hand in their software dev instances and all workmates are invited to hack them and report security vulnerabilities. That’s a lot of fun, but also gets frustrating towards the end when you don’t make any progress. :-) There’s also some actual hands-on training in advance for preparation of the two days. Unfortunately, I missed the last event due to my own project being very stressful at the time.

When I had a Do What You Want Day I also show my direct teammates what I learned in the hopes of this being interesting to them as well. I’m the only one in my team using this opportunity, sadly.

I’m finding this very interesting… An evolved neural network that plays the game of tic-tac-toe and so far is a pretty decent player. Here is a visualization of it’s evolved “brain” that underwent GA (genetic algorithm) training with classification learning + self-play.

prologic@JamessMacStudio
Sun May 25 21:44:41
~/tmp/neurog
 (main) 130
$ go build ./cmd/ttt/... && ./ttt
Generation  27 | Fitness: 0.486111 | Nodes: 44  | Conns: 82

… experimenting with building and training a tic-tac-toe game, which evolves a. neural net that learn to paly the game against the best evolved champions 😅

Pinellas County Running - Long run: 8.00 miles, 00:09:07 average pace, 01:12:50 duration
last long run of the training plan. legs were heavy again today. first half hour had a nice sprinkling going on but that humidity crept right in afterwards. more tired than i should be and need to try and keep these next few runs light
#running

We went on a 14 kilometers long hike in the heat, only a few spots were in the shade, most of our trip was in the open fields with the sun beating down on us. We reapplied the sun blocker after about two hours or so. All in all it took us about three and a half hours before we reached our destination Besigheim.

Last time I was there it was rainy, now we had the exact opposite. After some yummy Chinese lunch we visited the old town. There’s some gorgeous timer framing to see. When kept in decent shape, it just looks so dang cool.

Since it was too hot, we rode back by train. Despite the heat and some sections near the roaring Autobahn, this was a nice hike. Would do it again. Only in colder weather, though. I certainly don’t wanna trade my comperatively larger (still nothing to other more rural areas), covering forests with the wide open fields and vineyards in summer. That’s for sure.

https://lyse.isobeef.org/wanderung-von-asperg-nach-besigheim-2025-05-01/

LLMs can see and hear without any training
Comments ⌘ Read more

The Rise and Fall and Rise Again of Lionel Trains (1997)
Comments ⌘ Read more

Tell HN: Warning Google trains on your data when using aistudio.google.com
Comments ⌘ Read more

France’s new high-speed train has Americans asking: Why can’t we have that?
Comments ⌘ Read more

Serious issues in Llama 4 training. I Have Submitted My Resignation to GenAI”
Comments ⌘ Read more

Show HN: OCR pipeline for ML training (tables, diagrams, math, multilingual)
Comments ⌘ Read more

yeah! I’ve passed the critical point at factorio… I managed to automate blue science, trains and oil production… now it’s a great fun again…

MS-DOS and Windows 3.11 still run train dashboards at German railway (Jan’24)
Comments ⌘ Read more

Short summary of Project2025 and Trump’s plans for the US:

• Abolish the Federal Reserve
  Why? To end what is seen as an unelected, centralized body that exerts too much influence over the economy and monetary policy, replacing it with a more transparent, market-driven approach.

• Implement a national consumption tax
  Why? To replace the current federal income tax system, simplify taxation, and increase government revenue through a broader base that includes all consumers.

• Lower corporate tax rates
  Why? To promote business growth, increase investment, and stimulate job creation by reducing the financial burden on companies.

• Deregulate environmental policies
  Why? To reduce government intervention in the economy, particularly in energy and natural resources sectors, and to foster a more business-friendly environment.

• Restrict abortion access
  Why? To align with conservative pro-life values and overturn or limit abortion rights, seeking to restrict the practice at a federal level.

• Dismantle LGBTQ+ protections
  Why? To roll back protections viewed as promoting LGBTQ+ rights in areas like employment and education, in line with traditional family values.

• Eliminate diversity, equity, and inclusion (DEI) programs
  Why? To end policies that are seen as divisive and to promote a merit-based system that prioritizes individual achievements over group identity.

• Enforce stricter immigration policies, including mass deportations and detentions
  Why? To prioritize border security, reduce illegal immigration, and enforce existing laws more aggressively, as part of a broader strategy to safeguard U.S. sovereignty.

• Eliminate the Department of Education
  Why? To reduce federal control over education and shift responsibilities back to local governments and private sectors, arguing that education decisions should be made closer to the community level.

• Restructure the Department of Justice
  Why? To ensure the department aligns more closely with the administration’s priorities, potentially reducing its scope or focus on areas like civil rights in favor of law-and-order policies.

• Appoint political loyalists to key federal positions
  Why? To ensure that government agencies are headed by individuals who are committed to advancing the administration’s policies, and to reduce the influence of career bureaucrats.

• Develop training programs for appointees to execute reforms effectively
  Why? To ensure that political appointees are equipped with the knowledge and skills necessary to implement the proposed changes quickly and effectively.

• Provide a 180-day transition plan with immediate executive orders
  Why? To ensure that the incoming administration can swiftly implement its agenda and make major changes early in its term without delay.

Do y’all agree with any/all/some of these poliices? Hmmm 🤔

#Project2025 #US #Trump

@prologic@twtxt.net That boycott didn’t last very long, eh!?

Yeah, sounds like another hype train arriving at the station.

OpenAI doesn’t like it when you use “their” generated slop without permission
OpenAI says it has found evidence that Chinese artificial intelligence start-up DeepSeek used the US company’s proprietary models to train its own open-source competitor, as concerns grow over a potential breach of intellectual property. ↫ Cristina Criddle and Eleanor Olcott for the FT This is more ironic than writing a song called Ironic that lists situations that aren’t actually … ⌘ Read more

OpenAI Says It Has Evidence DeepSeek Used Its Model To Train Competitor
OpenAI says it has evidence suggesting Chinese AI startup DeepSeek used its proprietary models to train a competing open-source system through “distillation,” a technique where smaller models learn from larger ones’ outputs.

The San Francisco-based company, along with partner Microsoft, blocked suspected DeepSeek accounts from accessing … ⌘ Read more

Pinellas County Cycling: 7.75 miles, 00:05:28 average pace, 00:42:22 duration
cross training
#cycling

Pinellas County Running: 5.03 miles, 00:09:19 average pace, 00:46:55 duration
easy five to complete the first week of marathon training. my left glute/hip had some twinges
#running

Pinellas County Running: 3.14 miles, 00:09:12 average pace, 00:28:55 duration
third day of training with just a base run. definitely feeling the squats and lunges i did two days ago!
#running

Pinellas County Running: 3.12 miles, 00:08:53 average pace, 00:27:44 duration
first day of marathon training. i do not have a race in mind but wanted to start training again. was going too fast but 48F and just feeling overall good.
#running

Bluesky’s Open API Means Anyone Can Scrape Your Data for AI Training. It’s All Public
Bluesky says it will never train generative AI on its users’ data. But despite that, “one million public Bluesky posts — complete with identifying user information — were crawled and then uploaded to AI company Hugging Face,” reports Mashable (citing an article by 404 Media).

“Shortly after the article’s p … ⌘ Read more

Easy run: 5.06 miles, 00:09:43 average pace, 00:49:08 duration
kept it easy. legs were a bit tired, but wanted to enjoy the weather. not sure if i should start training again for a marathon just in case or just wing it.
#running

I had to go to the office today and both train rides worked out just fine. Surprising!

OpenAI, Google, Anthropic admit they can’t scale up their chatbots any further

Once you’ve trained your large language model on the entire written output of humanity, where do you go?

https://pivot-to-ai.com/2024/11/14/openai-google-anthropic-admit-they-cant-scale-up-their-chatbots-any-further/

So we’re going to destroy the environment for AI slop that isn’t fit for purpose now and, if you believe the above post, never will be.

Portion of the modified Twitter TOS that goes into effect today (it’s on right now), as summarised (ironically) by Google’s Gemini:

“In simpler terms, this means that when you share your content (like text, images, or videos) on the service, you’re giving the company permission to use it in various ways. They can copy, modify, distribute, and even use it to train their AI models. This includes sharing your content with others and using it on other platforms. You won’t be paid for this, but using the service itself is considered enough compensation.”

I like this comment on Slashdot in the above link:

LLMs don’t have an understanding of anything. They can only regurgitate derivations of what they’ve been trained on and can’t apply that to something new in the same ways that humans or even other animals can. The models are just so large that the illusion is impressive.

So true.

hop, entraînement terminé, j’ai fait le plein d’énergie avant d’aller donner un sang de qualitté ^^ #EFS #dondusang. Le niveau 7 de la méthode #lafay est par contre trop longue, je n’ai pas assez de temps pour faire ça bien. Tant pis dans ce cas, retour à la n°6 et j’y ajouterai 1 exercice jusqu’à épuisement tiré au sort. Ou alors je ressort le #TRX, il faut qu eje trouve où l’accrocher. #sport #training

Insecure Robot Vacuums From Chinese Company Deebot Collect Photos and Audio to Train Their AI
Long-time Slashdot reader schwit1 shared this report from Australia’s public broadcaster ABC:

Ecovacs robot vacuums, which have been found to suffer from critical cybersecurity flaws, are collecting photos, videos and voice recordings — taken inside customers’ houses — to train the company’ … ⌘ Read more

@bender@twtxt.net Ha! Maybe I should get on the Markdown train. You’re taking away my excuses.

Pinellas County - Zone 2: 5.86 miles, 00:13:01 average pace, 01:16:16 duration
just some “training” in the humidity and such trying to keep it in the blue zone. frustrating! felt fine though.
#running

@prologic@twtxt.net I don’t know what you mean when you call them stochastic parrots, or how you define understanding. It’s certainly true that current language models show an obvious lack of understanding in many situations, but I find the trend impressive. I would love to see someone achieve similar results with much less power or training data.

Regarding complexity budget, slow software, all that:

Very few people do take pride in building simple, elegant, high-quality systems, do they? Why is that? Why are huge shiny things with tons of features more attractive? 🤔

I never explicitly thought about this, to be honest. It was only at the back of my head. And I never tried to teach our younger “students” at work: “Hey, it’s a great achievement to build something simple and elegant. That’s something to be proud of!”

Worse, simple software is often described as “boring”. Yes, in a way, it is boring, because your brain doesn’t have to get into overdrive to understand it. But that’s exactly the point. And it’s hard to achieve that! Simple software isn’t just “fewer lines of code”, you have to be pretty clever to solve a problem in a simple and elegant way. So it’s something to be proud of.

Could this be an intuitive, emotional way to get more people on board the “simple software”-train? 🤔

Pinellas County Running: 3.16 miles, 00:08:30 average pace, 00:26:52 duration
aiming for whatever felt easy. the humidity really was heavy with a light fog. woke up with no real pain and it was not until the end of the run where i felt a slightly sharp pain around my left glute and a bit in the left hip as well. thinking i need to reduce mileage a bit and try to train around it until i feel good enough to get back in to a routine again.
#running

Here comes the train, through the twigs while the birds wait

Hello to the train pulling through the park on its way to the grain mill.

My email is such a cluster of noise. The only time i actually use it is to find out I have to do my security training or something. All communication is slack now days.

@movq@www.uninformativ.de Haha! yeah sounds about like my HS CS program. A math teacher taught visual basic and pascal. and over on the other end of the school we had “electronics” which was a room next to the auto body class where they had a bunch of random computer parts scavenged from the district decommissioned surplus storage.

The advanced class would piece together training kits for the basic class to put together.

Pinellas County - Long Run: 8.50 miles, 00:08:53 average pace, 01:15:32 duration
nice run in the rain. spoke to a soldier humping his way through the park in preparation for special forces training. overall run was great, but i did land funny on my left leg which made it sore for the next day or so.
#running

St Petersburg Distance Classic Marathon: 26.41 miles, 00:11:23 average pace, 05:00:39 duration
first marathon down. everything that could go wrong did. honestly i am just proud i did not quit. now i have to look at the run and figure out what i can tweak or add to my training. had a cramp start in my right quad at around mile 15. then around mile 18 both of my calves started to feel odd as if someone was lightly strumming my tendons. then they seized! this continued for the remainder of the marathon where i would walk then try to run and then stop when i had to. then during the entirety of the pace my nose would not stop dripping making it difficult to breathe. ha! also my shorts almost came down twice and i had to re-tie them while carrying my handheld water in my teeth. seriously, so many things i did not expect and had not happened in any previous runs.

really happy to be able to eat spicy food and have some alcoholic beverages again though!
#running #race