Op de veerboot van Terschelling naar Harlingen is een groep meisjes van ongeveer vijftien jaar bij ons komen zitten .

Je krijgt een paar cijfers cadeau, maar het grid van 9x9 moet foutloos ingevuld worden.

Een Precies Vier bestaat uit 16 woorden, begrippen of namen, die moeten worden verdeeld in precies vier groepen van vier. Er is telkens maar één oplossing mogelijk. Welke woorden vormen een connectie?

Een kruiswoordpuzzel, maar dan heel klein (en snel).

Als je wel zin hebt om te sudokuen, maar het liever bij een gridje van 5x5 houdt.

A hacker who breached Suno reportedly revealed source code and training-library details showing the AI music generator scraped millions of songs and lyrics from sources including YouTube Music, Deezer, Genius, Pond5, Jamendo, Freesound, and podcast RSS feeds. "The hacked data is a rare look at exactly how AI models and tools are built," reports 404 Media. "Suno is one of the largest AI music generation tools on the internet, and has been the subject of several major lawsuits from the record industry, which accused the company of training on millions of copyrighted songs." Suno maintains that its models were trained on publicly available music files and metadata as fair use. 404 Media reports: The Recording Industry Association of America accused Suno of ripping songs directly from YouTube; the hacked data seen by 404 Media confirms this. The hacked material includes source code that appears to be from 2023 and 2024 that includes scraping instructions and details about the scope of at least some of the scraping. For example, the comments in one file note that they will pull from "genius_hq, youtube_music, freesound, jamendo, imp, deezer, ytm_tagged," and that "non-music will be filtered out." A file called "youtube_music" notes that at the time the file was last updated, it had ingested "2,013,545 music clips." Another file contains comments about different datasets Suno had created, which included "113,879 hours of youtube_music," "17,615 hours of genius_hq," "410 hours of free sound," "19,514 hours of imslp," "3,726 hours of jamendo," "62,117 hours of pond5_music," "12,287 hours of deezer," "152,162 hours of ytm_tagged," and "103 hours of musescore_lyrics." In total, this is at least decades worth of music.

Other code the hacker shared with 404 Media appeared to look specifically for vocals by searching specifically for acapella versions of songs on YouTube. The code also suggested that Suno was using proxies to scrape songs from YouTube through a company called Bright Data, which sells scraping tools, infrastructure, and data services. Additional code shows that with the help of an online tool called PodcastIndex, Suno identified 420,000 different podcasts that had at least five, 30-minute episodes and sought to download roughly 1 million hours of podcasts.

[...] The hacker, ellie.191, told 404 Media they breached the company by hacking an individual employee using the Shai-Hulud worm, a supply chain attack that allowed hackers to harvest GitHub and cloud service credentials. They said they also accessed Suno's customer list, which included customers' emails and/or phone numbers and Stripe payment details, depending on what they used to login. The hacker provided a sample of some of the customers, some of whom confirmed to 404 Media they had used their phone number to sign up for Suno and said they were never notified of a breach. The hacker told 404 Media they had no specific motivation for hacking Suno and said "I like to hack anything and everything."

Read more of this story at Slashdot.

The FCC plans to vote on repealing local TV ownership limits, including the 39% national audience cap that currently restricts how much of the U.S. market a single broadcast group can reach. Engadget reports: On August 6, commissioners will hold a ballot to repeal Section 303 of the Communications Act, and with it the 39 percent rule. In essence, the rule limits the reach of a local TV network to no more than 39 percent of the U.S.' total audience market. In its place, the FCC would move to a system whereby it would personally approve or reject TV ownership deals on a case-by-case basis.

It's not clear if the FCC even has the authority to reject Section 303 without the explicit consent of the legislature. As Lawrence J. Spiwak wrote in the Yale Journal on Regulation back in January, Section 10 of the Communications Act expressly forbids the FCC from bending the rules around Section 303. "Americans no longer trust the legacy national media to report the news fairly or accurately," wrote FCC Chairman Brendan Carr in an op-ed published on Breitbart. "In fact, only eight percent of Americans have a great deal of trust in mass media. That figure is even lower among Republicans -- sitting at a mere three percent."

"... Many local broadcast TV stations are getting hollowed out as a result and turning into little more than mouthpieces for programming produced in New York and Hollywood," he alleged. "That is not what Congress or the FCC intended."

Read more of this story at Slashdot.

Weer voor Rotterdam 🌕 - 16-07-2026 01:16 CEST

In één oogopslag:
• 22.4°C · Licht bewolkt 🌕 | Min 20.1°C / Max 29.3°C

Verwachting voor vandaag:
• Min 20.1°C, Max 29.3°C (Bewolkt) ☁️, 🧭 1018.6 hPa ↘️ -2.6 hPa/24h, Windsnelheid: 22.3 km/u (6.2 m/s), richting: ↙ 43°

Uurlijkse voorspelling voor de komende 12 uur:

• 02:00: 22.0°C (Bewolkt) ☁️, 🧭 1021.2 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 13.7 km/u (3.8 m/s), richting: ↙ 44°
• 03:00: 21.5°C (Licht bewolkt) 🌕, 🧭 1021.0 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 13.3 km/u (3.7 m/s), richting: ↙ 38°
• 04:00: 21.0°C (Bewolkt) ☁️, 🧭 1020.8 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 13.7 km/u (3.8 m/s), richting: ↙ 34°
• 05:00: 21.1°C (Gedeeltelijk bewolkt) ☁️, 🧭 1020.5 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 12.2 km/u (3.4 m/s), richting: ↙ 33°
• 06:00: 21.1°C (Licht bewolkt) 🌤️, 🧭 1020.4 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 12.6 km/u (3.5 m/s), richting: ↙ 30°
• 07:00: 21.0°C (Licht bewolkt) 🌤️, 🧭 1020.3 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 11.9 km/u (3.3 m/s), richting: ↙ 33°
• 08:00: 21.7°C (Zonnig) ☀️, 🧭 1020.2 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 9.7 km/u (2.7 m/s), richting: ↙ 35°
• 09:00: 22.9°C (Zonnig) ☀️, Kans op neerslag 1%, 🧭 1020.2 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 10.1 km/u (2.8 m/s), richting: ↙ 23°
• 10:00: 23.9°C (Gedeeltelijk bewolkt) ⛅, Kans op neerslag 3%, 🧭 1020.0 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 8.6 km/u (2.4 m/s), richting: ↓ 21°
• 11:00: 25.1°C (Gedeeltelijk bewolkt) ⛅, Kans op neerslag 4%, 🧭 1019.8 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 8.6 km/u (2.4 m/s), richting: ↙ 28°
• 12:00: 25.6°C (Bewolkt) ☁️, Kans op neerslag 3%, 🧭 1019.5 hPa ➡️ 0.0 hPa/1h, Windsnelheid: 7.9 km/u (2.2 m/s), richting: ↙ 46°
• 13:00: 26.1°C (Bewolkt) ☁️, 🧭 1018.8 hPa ↘️ -0.7 hPa/1h, Windsnelheid: 5.8 km/u (1.6 m/s), richting: ↙ 41°

Voorspelling voor de komende dagen:

• vrijdag 17 juli: Min 20.6°C, Max 29.1°C (Bewolkt) ☁️, Kans op neerslag 4%, 🧭 1014.7 hPa ↘️ -3.9 hPa/24h, Windsnelheid: 21.6 km/u (6.0 m/s), richting: ↙ 42°
• zaterdag 18 juli: Min 18.7°C, Max 24.7°C (Bewolkt) ☁️, Kans op neerslag 3%, 🧭 1016.8 hPa ↗️ +2.1 hPa/24h, Windsnelheid: 19.1 km/u (5.3 m/s), richting: ↘ 331°
• zondag 19 juli: Min 16.3°C, Max 20.3°C (Lichte motregen) 🌦️, Neerslag 0.3 mm, Kans op neerslag 1%, 🧭 1019.8 hPa ↗️ +3.0 hPa/24h, Windsnelheid: 17.6 km/u (4.9 m/s), richting: ↘ 322°
• maandag 20 juli: Min 15.3°C, Max 20.0°C (Bewolkt) ☁️, Kans op neerslag 6%, 🧭 1022.4 hPa ↗️ +2.6 hPa/24h, Windsnelheid: 18.1 km/u (5.0 m/s), richting: ↘ 332°
• dinsdag 21 juli: Min 14.7°C, Max 21.0°C (Bewolkt) ☁️, Kans op neerslag 2%, 🧭 1024.5 hPa ↗️ +2.1 hPa/24h, Windsnelheid: 15.6 km/u (4.3 m/s), richting: ↘ 336°
• woensdag 22 juli: Min 14.0°C, Max 20.6°C (Lichte motregen) 🌦️, Neerslag 0.6 mm, Kans op neerslag 9%, 🧭 1024.0 hPa ↘️ -0.5 hPa/24h, Windsnelheid: 16.3 km/u (4.5 m/s), richting: ↘ 328°

Details:
• 🌡️ Huidige temperatuur (om 01:15): 22.4°C (Licht bewolkt)
• 🤚 Gevoelstemperatuur: 20.9°C (-1.5°C)
• 💨 Windsnelheid: 13.3 km/u (3.7 m/s), richting: ↙ 46°
• 🌬️ Windstoten: 25.6 km/h (7.1 m/s)
• 💧 Luchtvochtigheid: 57%
• 🧭 Luchtdruk: 1021.2 hPa ➡️ 0.0 hPa/3h
• 👁️ Zichtbaarheid: 27.0 km
• ☀️ UV-index: 0.0
• 🌅 Zonsopgang: 05:40 · 🌇 Zonsondergang: 21:55

Luchtkwaliteit:
• AQI: 42 🟢 (Goed)
• PM2.5: 7.9 μg/m³
• PM10: 11.4 μg/m³

Gegevens geleverd door Open-Meteo

#Rotterdam #Meteo #MeteoBot #MeteoNetherlands #OpenMeteo #FediMeteo

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

Thomas Hawk posted a photo:

A few highlights from a collection of lovely woodblock-printed sugoroku game boards from Japan. “The gameboards use stories and maps to teach children history and geography.”

Yet another 2020's food controversy. Currently the RDA of protein is 0.8 g/KG of body weight. Some say it is too low, especially if active.Others say that most Americans are getting adequate amounts of protein, but specific needs may vary from individual to individual. And those pushing high intake may have ulterior motives. Then there's the plant vs. animal sources debate. One side says we need animal protein for optimal absorption. Then there's another camp that says plant sources of protein are just fine. This guy (who has a PHD in nutrition) says animal sources are ideal, but vegan sources are okay, though there are nuances. Last, we have a doctor who says IGF-1 in animal sources can be potentially harmful to our health.

How AI will change the way scientific computing is done remains an open question. One relies on ultra-precise double-precision mathematics, while the other is perfectly happy working with 4 bits. On the surface, the two are diametrically opposed, two extremes of a spectrum we call high-performance computing (HPC) — and yes, whether you like it or not, AI is HPC. However, the latest AI offering from Cadence Design Systems, one of the biggest names in industrial HPC, offers a glimpse of how high- and low-precision compute could not just coexist, but work together to solve bigger and more complex problems faster and with fewer resources. Announced on Wednesday, Cadence’s AuraStack is an agentic AI system built to assist electrical engineers to design and test printed circuit boards (PCBs), or conduct advanced packaging design and testing — two tasks that have historically relied on highly precise simulations. AI is definitely a big piece of what Cadence has built; however the company isn’t replacing these tools with hallucination-prone AI models. Instead, AuraStack is a bit like Anthropic’s Claude Code or OpenAI’s Codex, but rather than writing, compiling, debugging, and running C or Rust in a sandbox, Cadence’s latest agent is designed to orchestrate its existing test and simulation suites. “AI is amplifying the value of our engineering products and technologies,” Michael Jackson, CVP of Cadence’s system design and analysis division, told The Register. In other words, the AI model — we’re told AuraStack integrates with a wide range of open and proprietary models — functions as a natural language interface capable of planning and orchestrating complex multi-step circuit design and testing workflows that run at higher precision using CPUs, GPUs, and other accelerators. “For example, if I'm going to check and fix the IR reliability, I need to identify the power management components. I need to create a simulation-ready power tree, and then I need to do the simulation, and then I need to provide feedback to the designer,” Jackson said. Cadence's existing product stack already automates many of these processes. The problem, Jackson explains, is that a PCB or package design often requires completing thousands of tasks throughout its development. “Sixty-five percent of an engineer's day is spent navigating and dealing with a lot of these tasks,” he said. By orchestrating that scutwork, Jackson claims that AuraStack can deliver a 15x boost to productivity by letting the designer focus on design and engineering decisions rather than the individual tasks. These gains are enough that several large players in the electronics space, including Nvidia, have already signed up for the service. Cadence isn’t just melding AI with HPC for chip design or advanced packaging. The engineering software provider has built similar agents for digital and analog chip design. The idea of using low precision compute to run AI models that orchestrate more precise single- and double-precision physics simulations isn’t new. Nvidia is one of the biggest champions of this approach, which makes sense seeing as its GPUs aren’t limited to training and running AI models, even if that’s what most folks are buying them for these days. Earlier this year, we explored how researchers at the Department of Energy’s Sandia National Laboratories used AI agents to develop and test new hypotheses. They described the system as a self-driving lab. However, those tests, while similar in concept to what Cadence is doing with AuraStack, didn’t use LLMs and instead used more mature architectures like variational auto-encoders. But given the success of code assistants, it’s not hard to imagine agent harnesses similar to AuraStack being used to automate lab equipment, perform simulations, and then iterate on the results, enabling scientists to continue their research even after they’ve nodded off for the night. ®