Forgot your password?
Close
typodupeerror

Comment Re:So much for the rule of law (Score 1) 79

Generational titles are always meant to be derogatory and never to uplift. You never year, "those amazing boomers." You also don't attack my argument, you ad hominem attack me (via my generation, UID or whatever) without addressing the argument. You seem to be a simpleton yourself as you can't argue and make silly reductionist statements without offering any real insight of value.
AI

Google Announces Gemma 4 Open AI Models, Switches To Apache 2.0 License 3

Posted by BeauHD from the new-and-improved dept.
An anonymous reader quotes a report from Ars Technica: Google's Gemini AI models have improved by leaps and bounds over the past year, but you can only use Gemini on Google's terms. The company's Gemma open-weight models have provided more freedom, but Gemma 3, which launched over a year ago, is getting a bit long in the tooth. Starting today, developers can start working with Gemma 4, which comes in four sizes optimized for local usage. Google has also acknowledged developer frustrations with AI licensing, so it's dumping the custom Gemma license.

Like past versions of its open-weight models, Google has designed Gemma 4 to be usable on local machines. That can mean plenty of things, of course. The two large Gemma variants, 26B Mixture of Experts and 31B Dense, are designed to run unquantized in bfloat16 format on a single 80GB Nvidia H100 GPU. Granted, that's a $20,000 AI accelerator, but it's still local hardware. If quantized to run at lower precision, these big models will fit on consumer GPUs. Google also claims it has focused on reducing latency to really take advantage of Gemma's local processing. The 26B Mixture of Experts model activates only 3.8 billion of its 26 billion parameters in inference mode, giving it much higher tokens-per-second than similarly sized models. Meanwhile, 31B Dense is more about quality than speed, but Google expects developers to fine-tune it for specific uses.

The other two Gemma 4 models, Effective 2B (E2B) and Effective 4B (E4B), are aimed at mobile devices. These options were designed to maintain low memory usage during inference, running at an effective 2 billion or 4 billion parameters. Google says the Pixel team worked closely with Qualcomm and MediaTek to optimize these models for devices like smartphones, Raspberry Pi, and Jetson Nano. Not only do they use less memory and battery than Gemma 3, but Google also touts "near-zero latency" this time around. The Apache 2.0 license is much more flexible with its terms of use for commercial restrictions, "granting you complete control over your data, infrastructure, and models," says Google.

Clement Delangue, co-founder and CEO of Hugging Face, called it "a huge milestone" that will help developers use Gemma for more projects and expand what Google calls the "Gemmaverse."

Comment Re:human vs slop (Score 0) 53

Facebook is tightly tied to the government, launched the day after DARPA shut down LifeLog and was originally funded by Peter Thiel. It's always been intended as a global surveillance system. OpenAI also has ties to the US government and any of the same Peter Thiel backed entities.

OpenAI benefits from a global control grid. You know what China has with surveillance and companies providing individual person credit scores? The US and EU governments want that, but automated and on steroids. These people are evil and OpenAI/Anthropic are devouring the world's data, eating through RAM and storage and ultimately push for technocratic subjugation of people to fit into their perfect regime.

Comment Re:AI can help here (Score 3, Insightful) 68

by SumDog (#66072880) Attached to: Sweden Swaps Screens For Books In the Classroom
Chromebooks had zero to do with education. They were 100% about Google forcing every high school student to have a Google account as early as possible. I bet less than 1% of parents said, "No, we're not doing that. Here's a Ubuntu laptop instead. Never sign in to Google, cause I said so."

Everyone wants digital tracking of every human: governments (via ID programs), Google, Meta (both are pretty much governments at this point), Anthropic, OpenAI ... they all want to know exactly who everyone is. They all want a Technocracy.

Comment Re:hohoho (Score 4, Interesting) 69

The article is paywalled and every other article I found was obviously LLM generated shit and didn't link to this new implementation. It took me a bit, but I found at least one of the Rust implementations of Claude's CLI:

https://github.com/Outcomefocu...

I was to see Anthropic choke on this so bad.

Courts still haven't really ruled on AI generated code in any big countries yet, as far as I can tell. Courts could view AI code the same as AI generated images: non-copyrightable. Generated images can still be subject to trademark if you try to commercialize them, but code not so much. If code ever gets rules as non-copyrightable, any generated code is open game if it gets leaked. Courts could also rule it is subject to copyright of the original training data holders.

Both of these outcomes would be equally devastating to the entire industry in entirely different ways. I'm kinda read to see it all burn.

Comment Re:Impossible to prevent (Score 3, Insightful) 27

by SumDog (#66070956) Attached to: Australia Readies Social Media Court Action Citing Teen Ban Breaches
Because it has NOTHING at all to do with protecting children. Getting around the bans is the point. It means they have to lock things down harder. The next setup is the OS level government ID bullshit that has already been passed in Brazil.

The end goal is that nobody can go online and view or interact with anything without a government ID token being transmitted. It's small things like this, or even the stupid California/Colorado law. They seem useless and ineffective, and that's the point. So they'll add more, and more and eventually you'll need a government CA to access anything and bam, you're now China.

Comment Re:Sony makes memory cards? (Score 1) 50

I still have a Pioneer head unit in my 2007 WRX. Why the fuck would anyone buy a modern car? You gotta rip out the dash and add a wiring harness to disable the damn permanent modem. Older cars are really where it's at. Get something old and restore it. It'll be better and last longer, and you can install a modern head unit and get a backup camera too.

I still use Sony cellphones too. Great cameras, and you can still use their nice camera apps after you de-Google the phone:

https://battlepenguin.com/tech...

Comment Re:What good is AI (Score 2) 50

There's not a lot of DRAM or NVRAM in a Chromebook, or even that new super cheap $500 Macbook. Be prepared for having near zero power on home compute and be required to send everything to Microsoft/Google/Amazon for the most basic shit. Need to edit video? Congratulations, we have video editing as a service! Enjoy scrubbing through 4k video with our shitty web interface, laggy UI and layers of Electron cancer. Upload straight to Youtube, and literally nowhere else. It's not like you have the storage to preserve your own copy anyway. Oh it got censored and deleted? Well too you shouldn't have been wrong-thinking or wrong-speaking. This will help you forget.

Slashdot Top Deals

Take care of the luxuries and the necessities will take care of themselves. -- Lazarus Long

Close