Linux fréttir

An anonymous reader quotes a report from Ars Technica: Google has announced a tiny version of its Gemma open model designed to run on local devices. Google says the new Gemma 3 270M can be tuned in a snap and maintains robust performance despite its small footprint. [...] Running an AI model locally has numerous benefits, including enhanced privacy and lower latency. Gemma 3 270M was designed with these kinds of use cases in mind. In testing with a Pixel 9 Pro, the new Gemma was able to run 25 conversations on the Tensor G4 chip and use just 0.75 percent of the device's battery. That makes it by far the most efficient Gemma model. Developers shouldn't expect the same performance level of a multi-billion-parameter model, but Gemma 3 270M has its uses. Google used the IFEval benchmark, which tests a model's ability to follow instructions, to show that its new model punches above its weight. Gemma 3 270M hits a score of 51.2 percent in this test, which is higher than other lightweight models that have more parameters. The new Gemma falls predictably short of 1 billion-plus models like Llama 3.2, but it gets closer than you might think for having just a fraction of the parameters. Google claims Gemma 3 270M is good at following instructions out of the box, but it expects developers to fine-tune the model for their specific use cases. Due to the small parameter count, that process is fast and low-cost, too. Google sees the new Gemma being used for tasks like text classification and data analysis, which it can accomplish quickly and without heavy computing requirements. You can download the new Gemma for free, and the model weights are available. There's no separate commercial licensing agreement, so developers can modify, publish, and deploy Gemma 3 270M derivatives in their tools. You can download Gemma 3 270M from Hugging Face and Kaggle in both pre-trained and instruction-tuned versions.

Read more of this story at Slashdot.

Some custom malware, some legit software tools

At least a dozen ransomware gangs have incorporated kernel-level EDR killers into their malware arsenal, allowing them to bypass almost every major endpoint security tool on the market, escalate privileges, and ultimately steal and encrypt data before extorting victims into paying a ransom.…

Apple has reintroduced blood oxygen monitoring to certain Apple Watch models in the U.S. by shifting the feature's calculations to the paired iPhone, sidestepping an ITC import ban stemming from its legal dispute with medical device maker Masimo. TechCrunch reports: Blood oxygen data will be measured and calculated on the user's paired iPhone, and results can be viewed in the Respiratory section of the Health app. This means users won't be able to view the data on their Apple Watch, as they'll need to do so on their iPhone. Apple says the update announced today is enabled by a recent U.S. Customs ruling, which means that the tech giant is allowed to import Apple Watches with the redesigned Blood Oxygen feature. The change doesn't affect previously sold models with the original version of the feature or units bought outside the U.S. The redesigned feature only applies to Apple Watches that were sold after the ITC import ban took effect in early 2024. These users can access the redesigned Blood Oxygen feature through an iPhone and Apple Watch software update coming on Thursday.

Read more of this story at Slashdot.

PayPal payment processing has been unavailable for Steam purchases in most countries since early July 2025, Valve has confirmed, with functionality limited to transactions in U.S. dollars, Euros, British Pounds, Japanese Yen, Australian dollars, and Canadian dollars. In a statement to RockPaperShotgun, the company said one of PayPal's acquiring banks terminated all Steam transaction processing. Valve linked the bank's decision to previous Mastercard-related content restrictions. The disruption began in early July 2025 when PayPal notified Valve of the immediate termination, leaving millions of users in affected regions without PayPal access and no certain timeline for resolution.

Read more of this story at Slashdot.

An anonymous reader shares a report: Astronomers have observed the calamitous result of a star that picked the wrong dance partner. They have documented what appears to be a new type of supernova, as stellar explosions are known, that occurred when a massive star tried to swallow a black hole with which it had engaged in a lengthy pas de deux. The star, which was at least 10 times as massive as our sun, and the black hole, which had a similar mass, were gravitationally bound to one another in what is called a binary system. But as the distance separating them gradually narrowed, the black hole's immense gravitational pull appears to have distorted the star -- stretching it out from its spherical shape -- and siphoned off material before causing it to explode. An AI algorithm detected the event in real time, enabling astronomers to conduct comprehensive observations. Data from four years before the supernova showed bright emissions as the black hole consumed its companion's outer hydrogen layer. The exact mechanism remains uncertain -- either gravitational distortion triggered the star's collapse or the black hole completely tore it apart first. Following the explosion, the black hole consumed residual stellar debris, growing more massive.

Read more of this story at Slashdot.