Meta debuted a new version of its powerful Llama AI model, its latest effort to keep pace with similar technology from companies like OpenAI, X and Google. The company describes Llama 3 8B and Llama 3 70B, containing 8 billion and 70 billion parameters respectively, as a "major leap" in performance compared to their predecessors.

Meta claims that the Llama 3 models, trained on custom-built 24,000 GPU clusters, are among the best-performing generative AI models available for their respective parameter counts. The company supports this claim by citing the models' scores on popular AI benchmarks such as MMLU, ARC, and DROP, which attempt to measure knowledge, skill acquisition, and reasoning abilities. Despite the ongoing debate about the usefulness and validity of these benchmarks, they remain one of the few standardized methods for evaluating AI models. Llama 3 8B outperforms other open-source models like Mistral's Mistral 7B and Google's Gemma 7B on at least nine benchmarks, showcasing its potential in various domains such as biology, physics, chemistry, mathematics, and commonsense reasoning.

TechCrunch adds: Now, Mistral 7B and Gemma 7B aren't exactly on the bleeding edge (Mistral 7B was released last September), and in a few of benchmarks Meta cites, Llama 3 8B scores only a few percentage points higher than either. But Meta also makes the claim that the larger-parameter-count Llama 3 model, Llama 3 70B, is competitive with flagship generative AI models including Gemini 1.5 Pro, the latest in Google's Gemini series.

AltStore PAL has become one of the first alternative app marketplaces to launch in the European Union. Developed by Riley Testut, AltStore PAL is marketed as an open-source project designed to distribute apps from independent developers. MacRumors reports: At launch, it features two apps, including Testut's Delta game emulator and clipboard manager app Clip. Delta is also being simultaneously released in the App Store outside of the European Union, but it looks like EU customers will need to download it from AltStore. Testut says that once AltStore PAL is "running smoothly," third-party app developers will be able to submit their apps for distribution outside of the App Store. The app marketplace is designed to be decentralized with no directory, so developers will need to self-promote their apps and direct users to their websites to install an app through AltStore.

Distributing apps through AltStore is free of charge, but it is worth noting that apps that see more than one million first annual installs will need to pay Apple an 0.50 euro Core Technology Fee. App marketplaces have to pay the fee for every install with no free allowance, so AltStore is charged 0.50 euros each time it is installed. To afford the fee, Testut is charging 1.50 euros per year for AltStore PAL access. Testut has been working on AltStore PAL since Apple announced plans to support alternative app marketplaces in iOS 17.4. It is open to all apps, but Testut says that it makes the most sense for "smaller, indie apps that otherwise couldn't exist due to App Store rules." AltStore PAL is equipped with Patreon integration to allow developers to monetize their apps. Developers can offer their apps to just their patrons, and this method of distribution also allows for a sub-1 million cap on those who can subscribe to use an app.

Microsoft's Beijing-based research group published a new open source AI model on Tuesday, only to remove it from the internet hours later after the company realized that the model hadn't gone through adequate safety testing. From a report: The team that published the model, which is comprised of China-based researchers in Microsoft Research Asia, said in a tweet on Tuesday that they "accidentally missed" the safety testing step that Microsoft requires before models can be published.

Microsoft's AI policies require that before any AI models can be published, they must be approved by the company's Deployment Safety Board, which tests whether the models can carry out harmful tasks such as creating violent or disturbing content, according to an employee familiar with the process. In a now-deleted blog post, the researchers behind the model, dubbed WizardLM-2, said that it could carry out tasks like generating text, suggesting code, translating between different languages, or solving some math problems.

The U.S. government is warning that smart locks securing entry to an estimated 50,000 dwellings nationwide contain hard-coded credentials that can be used to remotely open any of the locks. Krebs on SecurityL: The lock's maker Chirp Systems remains unresponsive, even though it was first notified about the critical weakness in March 2021. Meanwhile, Chirp's parent company, RealPage, Inc., is being sued by multiple U.S. states for allegedly colluding with landlords to illegally raise rents. On March 7, 2024, the U.S. Cybersecurity & Infrastructure Security Agency (CISA) warned about a remotely exploitable vulnerability with "low attack complexity" in Chirp Systems smart locks.

"Chirp Access improperly stores credentials within its source code, potentially exposing sensitive information to unauthorized access," CISA's alert warned, assigning the bug a CVSS (badness) rating of 9.1 (out of a possible 10). "Chirp Systems has not responded to requests to work with CISA to mitigate this vulnerability." Matt Brown, the researcher CISA credits with reporting the flaw, is a senior systems development engineer at Amazon Web Services. Brown said he discovered the weakness and reported it to Chirp in March 2021, after the company that manages his apartment building started using Chirp smart locks and told everyone to install Chirp's app to get in and out of their apartments.

An anonymous reader shares a report: Over the weekend, developer Mattia La Spina launched iGBA as one of the first retro game emulators legitimately available on the iOS App Store following Apple's rules change regarding such emulators earlier this month. As of Monday morning, though, iGBA has been pulled from the App Store following controversy over the unauthorized reuse of source code from a different emulator project.

iOS 8.1 plugs security hole that made it easy to install emulators Shortly after iGBA's launch, some people on social media began noticing that the project appeared to be based on the code for GBA4iOS, a nearly decade-old emulator that developer Riley Testut and a partner developed as high-schoolers (and distributed via a temporary security hole in the iOS App store). Testut took to social media Sunday morning to call iGBA a "knock-off" of GBA4iOS. "I did not give anyone permission to do this, yet it's now sitting at the top of the charts (despite being filled with ads + tracking)," he wrote.

GBA4iOS is an open source program released under the GNU GPLv2 license, with licensing terms that let anyone "use, modify, and distribute my original code for this project without fear of legal consequences." But those expansive licensing terms only apply "unless you plan to submit your app to Apple's App Store, in which case written permission from me is explicitly required."

The Linux Foundation-backed project OpenTofu "has gotten legal pushback from HashiCorp," according to a report — just seven months after forking OpenTofu's code from HashiCorp's IT deployment software Terraform: On April 3, HashiCorp issued a strongly-worded Cease and Desist letter to OpenTofu, accusing that the project has "repeatedly taken code HashiCorp provided only under the Business Software License (BSL) and used it in a manner that violates those license terms and HashiCorp's intellectual property rights." It goes on to note that "In at least some instances, OpenTofu has incorrectly re-labeled HashiCorp's code to make it appear as if it was made available by HashiCorp originally under a different license." Last August, HashiCorp announced that it would be transitioning its software from the open source Mozilla Public License (MPL 2.0) to the Business Source License (BSL), a license that permits the source to be viewed, but not run in production environments without explicit approval by the license owner. HashiCorp gave OpenTofu until April 10 to remove any allegedly copied code from the OpenTofu repository, threatening litigation if the project fails to do so.
Others are also covering the fracas, including Steven J. Vaughan-Nichols at DevOps.com: OpenTofu replied, "The OpenTofu team vehemently disagrees with any suggestion that it misappropriated, mis-sourced, or otherwise misused HashiCorp's BSL code. All such statements have zero basis in facts." In addition, it said, HashiCorp's claims of copyright infringement are completely unsubstantiated. As for the code in question, OpenTofu claims it can clearly be shown to have been copied from older code under the Mozilla Public License (MPL) 2.0. "HashiCorp seems to have copied the same code itself when they implemented their version of this feature. All of this is easily visible in our detailed SCO analysis, as well as their own comments."

In a detailed source code origination (SCO) examination of the problematic source code, OpenTofu stated that HashiCorp was mistaken. "We believe that this is just a case of a misunderstanding where the code came from." OpenTofu maintains the code was originally licensed under the MPL, not the BSL. If so, then OpenTofu was perfectly within its right to use the code in its codebase...

[OpenTofu's lawyer] concluded, "In the future, if you should have any concerns or questions about how source code in OpenTofu is developed, we would ask that you contact us first. Immediately issuing DMCA takedown notices and igniting salacious negative press articles is not the most helpful path to resolving concerns like this."

An anonymous reader quotes a report from The Guardian: Hospitals -- despite being places where people implicitly expect to have their personal details kept private -- frequently use tracking technologies on their websites to share user information with Google, Meta, data brokers, and other third parties, according to research published today. Academics at the University of Pennsylvania analyzed a nationally representative sample of 100 non-federal acute care hospitals -- essentially traditional hospitals with emergency departments -- and their findings were that 96 percent of their websites transmitted user data to third parties. Additionally, not all of these websites even had a privacy policy. And of the 71 percent that did, 56 percent disclosed specific third-party companies that could receive user information.

The researchers' latest work builds on a study they published a year ago of 3,747 US non-federal hospital websites. That found 98.6 percent tracked and transferred visitors' data to large tech and social media companies, advertising firms, and data brokers. To find the trackers on websites, the team checked out each hospitals' homepage on January 26 using webXray, an open source tool that detects third-party HTTP requests and matches them to the organizations receiving the data. They also recorded the number of third-party cookies per page. One name in particular stood out, in terms of who was receiving website visitors' information. "In every study we've done, in any part of the health system, Google, whose parent company is Alphabet, is on nearly every page, including hospitals," [Dr Ari Friedman, an assistant professor of emergency medicine at the University of Pennsylvania] observed. "From there, it declines," he continued. "Meta was on a little over half of hospital webpages, and the Meta Pixel is notable because it seems to be one of the grabbier entities out there in terms of tracking."

Both Meta and Google's tracking technologies have been the subject of criminal complaints and lawsuits over the years -- as have some healthcare companies that shared data with these and other advertisers. In addition, between 20 and 30 percent of the hospitals share data with Adobe, Friedman noted. "Everybody knows Adobe for PDFs. My understanding is they also have a tracking division within their ad division." Others include telecom and digital marketing companies like The Trade Desk and Verizon, plus tech giants Oracle, Microsoft, and Amazon, according to Friedman. Then there's also analytics firms including Hotjar and data brokers such as Acxiom. "And two thirds of hospital websites had some kind of data transfer to a third-party domain that we couldn't even identify," he added. Of the 71 hospital website privacy policies that the team found, 69 addressed the types of user information that was collected. The most common were IP addresses (80 percent), web browser name and version (75 percent), pages visited on the website (73 percent), and the website from which the user arrived (73 percent). Only 56 percent of these policies identified the third-party companies receiving user information. In lieu of any federal data privacy law in the U.S., Friedman recommends users protect their personal information via the browser-based tools Ghostery and Privacy Badger, which identify and block transfers to third-party domains.

An anonymous reader quotes a report from Ars Technica: Hardware sold for years by the likes of Intel and Lenovo contains a remotely exploitable vulnerability that will never be fixed. The cause: a supply chain snafu involving an open source software package and hardware from multiple manufacturers that directly or indirectly incorporated it into their products. Researchers from security firm Binarly have confirmed that the lapse has resulted in Intel, Lenovo, and Supermicro shipping server hardware that contains a vulnerability that can be exploited to reveal security-critical information. The researchers, however, went on to warn that any hardware that incorporates certain generations of baseboard management controllers made by Duluth, Georgia-based AMI or Taiwan-based AETN are also affected.

BMCs are tiny computers soldered into the motherboard of servers that allow cloud centers, and sometimes their customers, to streamline the remote management of vast fleets of servers. They enable administrators to remotely reinstall OSes, install and uninstall apps, and control just about every other aspect of the system -- even when it's turned off. BMCs provide what's known in the industry as "lights-out" system management. AMI and AETN are two of several makers of BMCs. For years, BMCs from multiple manufacturers have incorporated vulnerable versions of open source software known as lighttpd. Lighttpd is a fast, lightweight web server that's compatible with various hardware and software platforms. It's used in all kinds of wares, including in embedded devices like BMCs, to allow remote administrators to control servers remotely with HTTP requests. [...] "All these years, [the lighttpd vulnerability] was present inside the firmware and nobody cared to update one of the third-party components used to build this firmware image," Binarly researchers wrote Thursday. "This is another perfect example of inconsistencies in the firmware supply chain. A very outdated third-party component present in the latest version of firmware, creating additional risk for end users. Are there more systems that use the vulnerable version of lighttpd across the industry?"

The vulnerability makes it possible for hackers to identify memory addresses responsible for handling key functions. Operating systems take pains to randomize and conceal these locations so they can't be used in software exploits. By chaining an exploit for the lighttpd vulnerability with a separate vulnerability, hackers could defeat this standard protection, which is known as address space layout randomization. The chaining of two or more exploits has become a common feature of hacking attacks these days as software makers continue to add anti-exploitation protections to their code. Tracking the supply chain for multiple BMCs used in multiple server hardware is difficult. So far, Binarly has identified AMI's MegaRAC BMC as one of the vulnerable BMCs. The security firm has confirmed that the AMI BMC is contained in the Intel Server System M70KLP hardware. Information about BMCs from ATEN or hardware from Lenovo and Supermicro aren't available at the moment. The vulnerability is present in any hardware that uses lighttpd versions 1.4.35, 1.4.45, and 1.4.51. "A potential attacker can exploit this vulnerability in order to read memory of Lighttpd Web Server process," Binarly researchers wrote in an advisory. "This may lead to sensitive data exfiltration, such as memory addresses, which can be used to bypass security mechanisms such as ASLR." Advisories are available here, here, and here.

Liam Proven reports via The Register: Bad news for those who want to play with OpenVMS in non-production use. Older versions are disappearing, and the terms are getting much more restrictive. The corporation behind the continued development of OpenVMS, VMS Software, Inc. -- or VSI to its friends, if it has any left after this -- has announced the latest Updates to the Community Program. The news does not look good: you can't get the Alpha and Itanium versions any more, only a limited x86-64 edition.

OpenVMS is one of the granddaddies of big serious OSes. A direct descendant of the OSes that inspired DOS, CP/M, OS/2, and Windows, as well as the native OS of the hardware on which Unix first went 32-bit, VMS has been around for nearly half a century. For decades, its various owners have offered various flavors of "hobbyist program" under which you could get licenses to install and run it for free, as long as it wasn't in production use. Since Compaq acquired DEC, then HP acquired Compaq, its prospects looked checkered. HP officially killed it off in 2013, then in 2014 granted it a reprieve and sold it off instead. New owner VSI ported it to x86-64, releasing that new version 9.2 in 2022. Around this time last year, we covered VSI adding AMD support and opening a hobbyist program of its own. It seems from the latest announcement that it has been disappointed by the reception: "Despite our initial aspirations for robust community engagement, the reality has fallen short of our expectations. The level of participation in activities such as contributing open source software, creating wiki articles, and providing assistance on forums has not matched the scale of the program. As a result, we find ourselves at a crossroads, compelled to reassess and recalibrate our approach."

Although HPE stopped offering hobbyist licenses for the original VAX versions of OpenVMS in 2020, VSI continued to maintain OpenVMS 8 (in other words, the Alpha and Itanium editions) while it worked on version 9 for x86-64. VSI even offered a Student Edition, which included a freeware Alpha emulator and a copy of OpenVMS 8.4 to run inside it. Those licenses run out in 2025, and they won't be renewed. If you have vintage DEC Alpha or HP Integrity boxes with Itanic chips, you won't be able to get a legal licensed copy of OpenVMS for them, or renew the license of any existing installations -- unless you pay, of course. There will still be a Community license edition, but from now on it's x86-64 only. Although OpenVMS 9 mainly targets hypervisors anyway, it does support bare-metal operations on a single model of HPE server, the ProLiant DL380 Gen10. If you have one of them to play with -- well, tough. Now Community users only get a VM image, supplied as a VMWare .vmdk file. It contains a ready-to-go "OpenVMS system disk with OpenVMS, compilers and development tools installed." Its license runs for a year, after which you will get a fresh copy. This means you won't be able to configure your own system and keep it alive -- you'll have to recreate it, from scratch, annually. The only alternative for those with older systems is to apply to be an OpenVMS Ambassador.

Privacy startup Proton already offers an email app, a VPN tool, cloud storage, a password manager, and a calendar app. In April 2022, Proton acquired SimpleLogin, an open-source product that generates email aliases to protect inboxes from spam and phishing. Today, Proton acquired Standard Notes, advancing its already strong commitment to the open-source community. From a report: Standard Notes is an open-source note-taking app, available on both mobile and desktop platforms, with a user base of over 300,000. [...] Proton founder and CEO Andy Yen makes a point of stating that Standard Notes will remain open-source, will continue to undergo independent audits, will continue to develop new features and updates, and that prices for the app/service will not change. Standard Notes has three tiers: Free, which includes 100MB of storage, offline access, and unlimited device sync; Productivity for $90 per year, which includes features like markdown, spreadsheets with advanced formulas, Daily Notebooks, and two-factor authentication; and Professional for $120 per year, which includes 100GB of cloud storage, sharing for up to five accounts, no file limit size, and more.

Intel on Tuesday unveiled its new "Gaudi 3" AI chip that the company claims is over twice as power-efficient and can run AI models one-and-a-half times faster than Nvidia's H100 GPU. "It also comes in different configurations like a bundle of eight Gaudi 3 chips on one motherboard or a card that can slot into existing systems," adds CNBC. From the report: Intel tested the chip on models like Meta's open-source Llama and the Abu Dhabi-backed Falcon. It said Gaudi 3 can help train or deploy models, including Stable Diffusion or OpenAI's Whisper model for speech recognition. Intel says its chips use less power than Nvidia's. Intel said that the new Gaudi 3 chips would be available to customers in the third quarter, and companies including Dell, Hewlett Packard Enterprise, and Supermicro will build systems with the chips. Intel didn't provide a price range for Gaudi 3.

Gaudi 3 is built on a five nanometer process, a relatively recent manufacturing technique, suggesting that the company is using an outside foundry to manufacture the chips. In addition to designing Gaudi 3, Intel also plans to manufacture AI chips, potentially for outside companies, at a new Ohio factory expected to open in 2027 or 2028, CEO Patrick Gelsinger told reporters last month. "We do expect it to be highly competitive" with Nvidia's latest chips, said Das Kamhout, vice president of Xeon software at Intel, on a call with reporters. "From our competitive pricing, our distinctive open integrated network on chip, we're using industry-standard Ethernet. We believe it's a strong offering."

Meta Platforms is planning to launch two small versions of its forthcoming Llama 3 large-language model next week, The Information has reported [non-paywalled link]. From the report: The models will serve as a precursor to the launch of the biggest version of Llama 3, expected this summer. Release of the two small models will likely help spark excitement for the forthcoming Llama 3, which will be coming out roughly a year after Llama 2 launched last July.

It comes as several companies, including Google, Elon Musk's xAI and Mistral, have released open-source LLMs. Meta hopes Llama 3 will catch up with OpenAI's GPT-4, which can answer questions based on images users upload to the chatbot. The biggest version will be multimodal, which means it will be capable of understanding and generating both texts and images. In contrast, the two small models to be released next week won't be multimodal, the employee said.

The foundations behind Rust, Python, Apache, Eclipse, PHP, OpenSSL, and Blender announced plans to create "common specifications for secure software development," based on "existing open source best practices."

From the Eclipse Foundation: This collaborative effort will be hosted at the Brussels-based Eclipse Foundation [an international non-profit association] under the auspices of the Eclipse Foundation Specification Process and a new working group... Other code-hosting open source foundations, SMEs, industry players, and researchers are invited to join in as well.

The starting point for this highly technical standardisation effort will be today's existing security policies and procedures of the respective open source foundations, and similar documents describing best practices.

The governance of the working group will follow the Eclipse Foundation's usual member-led model but will be augmented by explicit representation from the open source community to ensure diversity and balance in decision-making. The deliverables will consist of one or more process specifications made available under a liberal specification copyright licence and a royalty-free patent licence... While open source communities and foundations generally adhere to and have historically established industry best practices around security, their approaches often lack alignment and comprehensive documentation.

The open source community and the broader software industry now share a common challenge: legislation has introduced an urgent need for cybersecurity process standards.
The Apache Foundation notes the working group is forming partly "to demonstrate our commitment to cooperation with and implementation of" the EU's Cyber Resilience Act. But the Eclipse Foundation adds that even before it goes into effect in 2027, they're recognizing open source software's "increasingly vital role in modern society" and an increasing need for reliability, safety, and security, so new regulations like the CRA "underscore the urgency for secure by design and robust supply chain security standards."

Their announcement adds that "It is also important to note that it is similarly necessary that these standards be developed in a manner that also includes the requirements of proprietary software development, large enterprises, vertical industries, and small and medium enterprises." But at the same time, "Today's global software infrastructure is over 80% open source... [W]hen we discuss the 'software supply chain,' we are primarily, but not exclusively, referring to open source."

"We invite you to join our collaborative effort to create specifications for secure open source development," their announcement concludes," promising initiative updates on a new mailing list. "Contribute your ideas and participate in the magic that unfolds when open source foundations, SMEs, industry leaders, and researchers combine forces to tackle big challenges."

The Python Foundation's announcement calls it a "community-driven initiative" that will have "a lasting impact on the future of cybersecurity and our shared open source communities."

AMD plans to document and open source its Micro Engine Scheduler (MES) firmware for GPUs, giving users more control over Radeon graphics cards. From a report: It's part of a larger effort AMD confirmed earlier this week about making its GPUs more open source at both a software level in respect to the ROCm stack for GPU programming and a hardware level. Details were scarce with this initial announcement, and the only concrete thing it introduced was a GitHub tracker.

However, yesterday AMD divulged more details, specifying that one of the things it would be making open source was the MES firmware for Radeon GPUs. AMD says it will be publishing documentation for MES around the end of May, and will then release the source code some time afterward. For one George Hotz and his startup, Tiny Corp, this is great news. Throughout March, Hotz had agitated for AMD to make MES open source in order to fix issues he was experiencing with his RX 7900 XTX-powered AI server box. He had talked several times to AMD representatives, and even the company's CEO, Lisa Su.

The Document Foundation: Following a successful pilot project, the northern German federal state of Schleswig-Holstein has decided to move from Microsoft Windows and Microsoft Office to Linux and LibreOffice (and other free and open source software) on the 30,000 PCs used in the local government. As reported on the homepage of the Minister-President: "Independent, sustainable, secure: Schleswig-Holstein will be a digital pioneer region and the first German state to introduce a digitally sovereign IT workplace in its state administration. With a cabinet decision to introduce the open-source software LibreOffice as the standard office solution across the board, the government has given the go-ahead for the first step towards complete digital sovereignty in the state, with further steps to follow."

The UK and US have signed a landmark deal to work together on testing advanced artificial intelligence (AI) and develop "robust" safety methods for AI tools and their underlying systems. "It is the first bilateral agreement of its kind," reports the BBC. From the report: UK tech minister Michelle Donelan said it is "the defining technology challenge of our generation." "We have always been clear that ensuring the safe development of AI is a shared global issue," she said. "Only by working together can we address the technology's risks head on and harness its enormous potential to help us all live easier and healthier lives."

The secretary of state for science, innovation and technology added that the agreement builds upon commitments made at the AI Safety Summit held in Bletchley Park in November 2023. The event, attended by AI bosses including OpenAI's Sam Altman, Google DeepMind's Demis Hassabis and tech billionaire Elon Musk, saw both the UK and US create AI Safety Institutes which aim to evaluate open and closed-source AI systems. [...]

Gina Raimondo, the US commerce secretary, said the agreement will give the governments a better understanding of AI systems, which will allow them to give better guidance. "It will accelerate both of our Institutes' work across the full spectrum of risks, whether to our national security or to our broader society," she said. "Our partnership makes clear that we aren't running away from these concerns - we're running at them."

An anonymous reader quotes a report from Reuters: The U.S. Federal Communications Commission will vote to reinstate landmark net neutrality rules and assume new regulatory oversight of broadband internet that was rescinded under former President Donald Trump, the agency's chair said. The FCC told advocates on Tuesday of the plan to vote on the final rule at its April 25 meeting. The commission voted 3-2 in October on the proposal to reinstate open internet rules adopted in 2015 and re-establish the commission's authority over broadband internet.

Net neutrality refers to the principle that internet service providers should enable access to all content and applications regardless of the source, and without favoring or blocking particular products or websites. FCC Chair Jessica Rosenworcel confirmed the planned commission vote in an interview with Reuters. "The pandemic made clear that broadband is an essential service, that every one of us -- no matter who we are or where we live -- needs it to have a fair shot at success in the digital age," she said. "An essential service requires oversight and in this case we are just putting back in place the rules that have already been court-approved that ensures that broadband access is fast, open and fair."

Lindsay Clark reports via The Register: Analytics platform Databricks has launched an open source foundational large language model, hoping enterprises will opt to use its tools to jump on the LLM bandwagon. The biz, founded around Apache Spark, published a slew of benchmarks claiming its general-purpose LLM -- dubbed DBRX -- beat open source rivals on language understanding, programming, and math. The developer also claimed it beat OpenAI's proprietary GPT-3.5 across the same measures.

DBRX was developed by Mosaic AI, which Databricks acquired for $1.3 billion, and trained on Nvidia DGX Cloud. Databricks claims it optimized DBRX for efficiency with what it calls a mixture-of-experts (MoE) architecture â" where multiple expert networks or learners divide up a problem. Databricks explained that the model possesses 132 billion parameters, but only 36 billion are active on any one input. Joel Minnick, Databricks marketing vice president, told The Register: "That is a big reason why the model is able to run as efficiently as it does, but also runs blazingly fast. In practical terms, if you use any kind of major chatbots that are out there today, you're probably used to waiting and watching the answer get generated. With DBRX it is near instantaneous."

But the performance of the model itself is not the point for Databricks. The biz is, after all, making DBRX available for free on GitHub and Hugging Face. Databricks is hoping customers use the model as the basis for their own LLMs. If that happens it might improve customer chatbots or internal question answering, while also showing how DBRX was built using Databricks's proprietary tools. Databricks put together the dataset from which DBRX was developed using Apache Spark and Databricks notebooks for data processing, Unity Catalog for data management and governance, and MLflow for experiment tracking.

"After taking a look at Floorp Browser, I was left wondering whether there was a Chromium-based web browser that was as good, or even better than Chrome," writes a "First Look" reviewer at It's Foss News.

"That is when I came across Thorium, a web-browser that claims to be the 'the fastest browser on Earth'." [Thorium] is backed by a myriad of tweaks that include, compiler optimizations for SSE4.2, AVS, AES, various mods to CFLAGS, LDFLAGS, thinLTO flags, and more. The developer shares performance stats using popular benchmarking tools... I tested it using Speedometer 3.0 benchmark on Fedora 39 and compared it to Brave, and the scores were:

Thorium: 19.2; Brave: 19.5

So, it may not be the "fastest" always, probably one of the fastest, that comes close to Brave or sometimes even beats it (depends on the version you tested it and your system).

Alexander Frick, the lead developer, also insists on providing support for older operating systems such as Windows 7 so that its user base can use a capable modern browser without much fuss... As Thorium is a cross-platform web browser, you can find packages for a wide range of platforms such as Linux, Raspberry Pi, Windows, Android, macOS, and more.
Thorium can sync to your Google account to import your bookmarks, extensions, and themes, according to the article.

"Overall, I can confidently say that it is a web browser I could daily drive, if I were to ditch Chrome completely. It gels in quite well with the Google ecosystem and has a familiar user interface that doesn't get in the way."

A Linux Foundation subsidiary has developed a free and open-source 3D game engine distributed under the Apache license. And last week the Open 3D Foundation announced "a big step forward, showcasing the power of open-source technologies in giving gamers around the globe unforgettable gaming experiences."

"We are proud to unveil MadWorld as the first mobile title powered by O3DE," said Joe Bryant, Executive Director of the Open 3D Foundation, "demonstrating the large potential of open-source technologies in game development."

And then this week Los Angeles Business Journal reported that El Segundo-based gaming studio Carbonated Inc. "has raised $11 million of series A funding to finance the development and release of its debut game title... Prior to its most recent round, Carbonated closed an $8.5 million seed funding round in 2020, which also included participation from Andreessen and Bitkraft." Since its founding [in 2015], the company has been focusing on research and development for its upcoming first title, called "MadWorld." The third-person, multiplayer shooter game is set in a post-apocalyptic world and features both player-versus-player and player-versus-environment features. Players of the game will battle for land control in a dystopian setting. Using a combination of open-source mapping tools and Carbonated's proprietary custom operations technology, called Carbyne, the game's world is designed around real-life cities and locations. Players are initially dropped into the game's version of their own real-time location.

The game allows players to optionally engage using blockchain technology with a digital asset-ownership layer powered by a blockchain network called XPLA.
Earlier this month Madworld "opened up for Early Access registration," reports the egamers web site, arguing that the game "is set to redefine the gaming landscape and will make its public debut later this year." After a catastrophic event named "The Collapse," MadWorld takes place in a desolate Earth where players engage in a battle for survival, highlighting the game's unique setting and immersive experience. The game's world is intricately designed with 250,000 land plots mapped out on a hexagonal grid, each presenting unique resources and strategic benefits. This innovative approach to game design enhances the gameplay experience and introduces a new layer of strategy and competition.

MadWorld's gameplay is centered around integrating Web3 technologies, which allows for the ownership, enhancement, and trading of tokenized representations of real-world locations. This feature encourages players to create clans and work together or compete for essential resources that are spread across the vast game world. Clans can acquire these resources by paying tributes to NFT landowners using "Rounds," the in-game currency. This mechanism not only fosters a sense of community and teamwork but also creates unique economic opportunities within the game by blending traditional gaming elements with the emerging field of digital assets.
"With its use of O3DE, Carbonated can enhance the game's visual fidelity, performance, and scalability," according to the Linux Foundation's announcement, "in order to deliver a fast-paced adventure on mobile platforms." O3DE is an open-source game engine developed by a collaborative community of industry experts. It includes state-of-the-art rendering capabilities, dynamic lighting, and realistic physics simulation. These features have enabled Carbonated to build realistic dystopian environments and create action-packed gameplay in MadWorld.
According to its official site, MadWorld "is set to be released to the public sometime in 2024 and is currently being tested on iOS and Android operating systems."

Carbonated's CEO Travis Boatman made this prediction to the site Decrypt. "We think mobile is where the breakout will happen for Web3."